Mercurial > hg > wm
view Meerwald/wavelet.c @ 18:3bdb67e76858
mse opt.
author | Peter Meerwald <pmeerw@cosy.sbg.ac.at> |
---|---|
date | Fri, 30 Jan 2009 12:46:49 +0100 |
parents | f83ef905a63d |
children | bd669312f068 |
line wrap: on
line source
#include <stdio.h> #include <stdlib.h> #include <string.h> #include <math.h> #include "wavelet.h" #include <ctype.h> #include <values.h> static int read_char(FILE *fp); static int read_int(FILE *fp); IntImage new_intimage(int width, int height) { IntImage image; image = (IntImage) calloc(1,sizeof(struct IntImage_struct)); if (image==NULL) goto error; image->data = (IntPixel*) calloc(width*height,sizeof(IntPixel)); if (image->data==NULL) goto error; image->width = width; image->height = height; image->size = width*height; image->bpp = 0; image->min_val = (IntPixel) 0; image->max_val = (IntPixel) 0; return image; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } Image new_image(int width, int height) { Image image; image = (Image) calloc(1,sizeof(struct Image_struct)); if (image==NULL) goto error; image->data = (Pixel*) calloc(width*height,sizeof(Pixel)); if (image->data==NULL) goto error; image->width = width; image->height = height; image->size = width*height; image->bpp = 0; image->min_val = (Pixel) 0; image->max_val = (Pixel) 0; return image; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } void free_intimage(IntImage img) { if (img) { if (img->data) free(img->data); free(img); } } void free_image(Image img) { if (img) { if (img->data) free(img->data); free(img); } } /************************************************************************ * Functionname: load_intimage * -------------------------------------------------------------------- * PARAMETER: * file: filename of image * max_val: scaling of grey values to [0..max_val] * * RETURN: * Image that shoud be loaded, if not possible return NULL * -------------------------------------------------------------------- * DESCRIPTION: * This function loads an IntImage (PGM, ASCII or binary * encoded (P5 or P3 format) ) from a file. ************************************************************************/ IntImage load_intimage(char *file, int max_val) { IntImage img; FILE *fp; IntPixel *data; int width, height, i, max, ch1, ch2; fp=fopen(file,"rb"); if (fp==NULL) goto error; ch1=getc(fp); ch2=getc(fp); if (ch1!='P' || (ch2!='5' && ch2!='2')) goto error1; width=read_int(fp); height=read_int(fp); if ((width==0) || (height==0) ) goto error1; max=read_int(fp); img=new_intimage(width,height); img->bpp=8; data=img->data; for (i=0; i<img->size; i++) { if (ch2=='5') *data=getc(fp); else *data=read_int(fp); data++; } fclose(fp); return img; error1: err_SimpleMessage(err_GetErrorMessage(Error_WrongFileFormat)); return NULL; error: err_SimpleMessage(err_GetErrorMessage(Error_CantOpenFile)); return NULL; } /************************************************************************ * Functionname: load_image * -------------------------------------------------------------------- * PARAMETER: * file: filename of image * max_val: scaling of grey values to [0..max_val] * * RETURN: * Image that shoud be loaded, if not possible return NULL * -------------------------------------------------------------------- * DESCRIPTION: * This function loads an IntImage with load_intimage * and then converts to Image. ************************************************************************/ Image load_image(char *file, int max_val) { Image img; IntImage intimg; intimg = load_intimage(file, max_val); if (!intimg) return NULL; img = intimage_to_image(intimg); if (!intimg) return NULL; return img; } /************************************************************************/ /* Functionname: save_image_P5 */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* img: Image that shoud be saved */ /* file: filename of image */ /* -------------------------------------------------------------------- */ /* Description: save an image as PGM (P5 binary decoded) file */ /* */ /************************************************************************/ int save_image_P5(char *file, Image img) { FILE *fp; Pixel *data; long i; int p; fp=fopen(file,"wb"); if (fp==NULL) goto error; fprintf(fp,"P5\n"); fprintf(fp,"%d %d\n%d ",img->width,img->height,255); data=img->data; for (i=0;i<img->size;i++) { p=floor(*data+0.5); if (p<0) p=0; if (p>255) p=255; /* putc(*data,fp); */ putc(p,fp); data++; } fclose(fp); return 1; error: err_SimpleMessage(err_GetErrorMessage(Error_CantOpenFile)); return 0; } void clear_image(Image img) { int i; PreCondition(img!=NULL,"image==NULL"); for (i=0;i<img->size;i++) (img->data)[i]=(Pixel) 0; } void copy_into_image(Image img1,Image img2,int x,int y) /* copy img2 into img1 at position (x,y)*/ { int start,i,j,aim; Pixel *temp; temp=img2->data; start=img1->width*y+x; for (i=0;i<img2->height;i++) { for (j=0;j<img2->width;j++) { aim=start+j+img1->width*i; img1->data[aim]=*temp; temp++; } } } void copy_into_intimage(IntImage img1,IntImage img2,int x,int y) {/* copy img2 into img1 at position (x,y)*/ int start,i,j,aim; IntPixel *temp; temp=img2->data; start=img1->width*y+x; for (i=0;i<img2->height;i++) { for (j=0;j<img2->width;j++) { aim=start+j+img1->width*i; img1->data[aim]=*temp; temp++; } } } void copy_part_of_image_into_image(Image dest_img, int dest_x, int dest_y, Image src_img, int src_x, int src_y, int width, int height) { Pixel *sp,*dp; int y,siz; sp=get_pixel_adr(src_img,src_x,src_y); dp=get_pixel_adr(dest_img,dest_x,dest_y); siz=width*sizeof(Pixel); for (y=0;y<height;y++) { memcpy(dp,sp,siz); sp+=src_img->width; dp+=dest_img->width; } } void copy_part_of_image(Image img1,Image img2,int x,int y) /* copy part of img2 begining at position (x,y) into img1 */ { int i,j,width,height,start,step; Pixel *data; width=img1->width; height=img1->height; start=img2->width*y+x; data=img1->data; for (i=0;i<height;i++) { step=i*img2->width; for (j=0;j<width;j++){ *data=img2->data[start+j+step]; data++; } } } void scale_image(Image img, int maximum) /* scale image to [0..maximum]*/ { int i; Pixel max = MINDOUBLE, min = MAXDOUBLE, multi; for (i=0;i<img->size;i++) { if (img->data[i]<min) min=img->data[i]; else if (img->data[i]>max) max=img->data[i]; } multi=(Pixel)maximum/(max-min); for (i=0;i<img->size;i++) img->data[i]=multi*(img->data[i]-min); img->min_val=0; img->max_val=maximum; } int string_to_pixel(char *str, Pixel *p) { float ppp; if (sscanf(str,"%f",&ppp)==1) { *p=(Pixel) ppp; return 1; } else { *p=0.0; return 0; } } Image intimage_to_image(IntImage i) { Image img; int j; img=new_image(i->width,i->height); if (img==NULL) goto error; for (j=0;j<i->size;j++) img->data[j]=(Pixel)i->data[j]; img->bpp=i->bpp; return img; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } IntImage image_to_intimage(Image i) { IntImage img; int j,multi=1,max,min; img=(IntImage)calloc(1,sizeof(struct IntImage_struct)); if (img==NULL) goto error; img->data=(IntPixel *)calloc(i->size,sizeof(IntPixel)); if (img->data==NULL) goto error; img->width=i->width; img->height=i->height; img->size=i->size; img->bpp=i->bpp; max=i->max_val; min=i->min_val; if ((max-min)!=0) multi=255.0/(max-min); for (j=0;j<img->size;j++) img->data[j]=(int)((i->data[j]-min)*multi+0.5); return img; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } static int read_char(FILE *fp) /*read a character from file, but skip any comments*/ { int ch; ch=getc(fp); if (ch=='#'){ do { ch=getc(fp); } while (ch!='\n' && ch!=EOF); } return ch; } static int read_int(FILE *fp) /*read an ascii integer from file, and skip leading tabstops,new lines ...*/ { int r,ch; do { ch=read_char(fp); } while (ch==' ' || ch=='\n' || ch=='\t'); if (ch<'0' || ch>'9') goto error; r=ch-'0'; while ( (ch=read_char(fp)) >='0' && (ch <= '9') ) { r*=10; r+=ch-'0'; } return r; error: return 0; } Image_tree new_image_tree() { Image_tree t; t=(Image_tree) calloc(1,sizeof(struct Image_tree_struct)); t->entropy=0.0; t->image=NULL; t->coarse=t->horizontal=t->vertical=t->diagonal=t->doubletree=NULL; t->level=0; t->codec_data=NULL; t->significance_map=NULL; return t; } void free_image_tree(Image_tree t) { if (t->coarse) free_image_tree(t->coarse); if (t->horizontal) free_image_tree(t->horizontal); if (t->vertical) free_image_tree(t->vertical); if (t->diagonal) free_image_tree(t->diagonal); if (t->doubletree) free_image_tree(t->doubletree); if (t->image) free_image(t->image); if (t->significance_map) free_intimage(t->significance_map); if (t->codec_data) free(t->codec_data); t->image=NULL; free(t); } /*********************************************************************** * Functionname: get_image_infos * -------------------------------------------------------------------- * Parameter: * Image image: input image * Pixel *min,*max,*avg,*var: return minimum, maximum, * average and variance of current image * -------------------------------------------------------------------- * Description: * get statistical information of Image ************************************************************************/ void get_image_infos(Image image, Image_info info) { int x,y; Pixel p,sp,sp2; sp=sp2=(Pixel)0.0; p=get_pixel(image,0,0); info->min=info->max=p; for (y=0;y<image->height;y++) for (x=0;x<image->width;x++) { p=get_pixel(image,x,y); info->max=MAX(info->max,p); info->min=MIN(info->min,p); sp+=p; sp2+=p*p; } sp=sp/image->width/image->height; sp2=sp2/image->width/image->height; info->mean=sp; info->var=sp2-sp*sp; info->rms=sqrt(sp2); } /*********************************************************************** * Functionname: get_difference_image * -------------------------------------------------------------------- * Parameter: * Image image1, image 2: input images * * Return: * Image : difference of image1 and image 2, * NULL if error occurs ************************************************************************/ Image get_difference_image(Image image1, Image image2) { Image diff; int i,max,w,h; Pixel *d,*i1,*i2; if ((!image1) || (!image2)) return NULL; w=image1->width; h=image1->height; if (image2->width != w || image2->height != h) return NULL; diff=new_image(w,h); max=w*h; d=diff->data; i1=image1->data; i2=image2->data; for (i=0;i<max;i++) d[i]=i2[i]-i1[i]; return diff; } /************************************************************************/ /* Functionname: get_intimage_infos */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* IntImage image: input image */ /* IntPixel *min,*max, return minimum, maximum */ /* Pixel *avg,*var: average and variance of current image */ /* average and variance of current image */ /* -------------------------------------------------------------------- */ /* Description: */ /* get statistical information of Image */ /************************************************************************/ void get_intimage_infos(IntImage image, IntPixel *min, IntPixel *max, Pixel *avg, Pixel *var) { int x,y; Pixel p,sp,sp2; sp=sp2=(Pixel)0.0; p= (Pixel) get_intpixel(image,0,0); *min=*max=p; for (y=0;y<image->height;y++) for (x=0;x<image->width;x++) { p= (Pixel) get_intpixel(image,x,y); *max=MAX(*max, (IntPixel) p); *min=MIN(*min, (IntPixel) p); sp+=p; sp2+=p*p; } sp=sp/image->width/image->height; sp2=sp2/image->width/image->height; *avg=sp; *var=sp2-sp*sp; } /************************************************************************/ /* Functionname: init_zigzag */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* Zigzag_data_struct: */ /* output: will be initialized, x/y hold coordinates of */ /* the first pixel */ /* int width,height: */ /* input: width/height of image: */ /* -------------------------------------------------------------------- */ /* Description: */ /* initializes Zigzag_data structure for use with next_zigzag */ /************************************************************************/ void init_zigzag(Zigzag_data zz, int width, int height) { zz->x=0; zz->y=0; zz->dir=zigzag_up; zz->w=width; zz->h=height; } /************************************************************************/ /* Functionname: next_zigzag */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* Zigzag_data_struct: */ /* int x,y: */ /* input: current position of zigzag-scan */ /* output: next position of zigzag-scan */ /* int w,h: width and height of image */ /* enum zigzag_direction *dir: i/o: */ /* direction moving thru the image */ /* -------------------------------------------------------------------- */ /* Description: */ /* calculates the next point (x',y') of the zigzag-scan */ /* through the image with size (w,h) */ /************************************************************************/ void next_zigzag(Zigzag_data zz) { switch(zz->dir) { case zigzag_up: if (zz->y==0) { if (zz->x==zz->w-1) { (zz->y)++; zz->dir=zigzag_down; } else { (zz->x)++; zz->dir=zigzag_down; } } else { if (zz->x==zz->w-1) { (zz->y)++; zz->dir=zigzag_down; } else { (zz->x)++; (zz->y)--; } } break; case zigzag_down: if (zz->x==0) { if (zz->y==zz->h-1) { (zz->x)++; zz->dir=zigzag_up; } else { (zz->y)++; zz->dir=zigzag_up; } } else { if (zz->y==zz->h-1) { (zz->x)++; zz->dir=zigzag_up; } else { (zz->x)--;(zz->y)++; } } break; } } Image get_absolute_image_scaled(Image img) { Image out; int x,y; struct Image_info_struct info; Pixel scale,p; out=new_image(img->width,img->height); get_image_infos(img, &info); scale=255/MAX(fabs(info.min),fabs(info.max)); for(y=0;y<img->height;y++) for(x=0;x<img->width;x++) { p=get_pixel(img,x,y)*scale; set_pixel(out,x,y,p); } return out; } #define FLOOR_HALF(x) ((x)&1 ? ((x)-1)/2 : (x)/2) #define CEILING_HALF(x) ((x)&1 ? ((x)+1)/2 : (x)/2) #define MOD(a,b) ( (a)<0 ? ((b)-((-(a))%(b))) : (a)%(b) ) Filter new_filter(int size) { Filter f; Entering; f=(Filter) calloc(1,sizeof(struct FilterStruct)); f->data=(Pixel *)calloc(size,sizeof(Pixel)); f->len=size; f->hipass=0; Leaving; return f; } Pixel get_filter_center(Filter f) { int i; Pixel p, sum, norm; if (f==NULL) return 0; sum=norm=0; for (i=0;i<f->len;i++) { p=f->data[i]; p=p*p; norm += p; sum += (i+f->start)*p; } p=sum/norm; return p; } int filter_cutoff(Image in, int in_start, int in_len, int in_step, Image out, int out_start, int out_len, int out_step, Filter f) { int i,i2,j; Pixel *out_pix, *in_pix, *f_data; int fstart,fend; /* Source interval */ Entering; PreCondition(out_len == in_len/2,"out_len != in_len/2 !!!"); /* convolution: out[i]=sum_{j=start}^{end} (in[2*i-j]*f[j]) boundaries: image in [in_start ... in_start + in_len-1] image out [out_start ... out_start + out_len-1] filter f [0..f->len-1] = [f->start .. f->end] cutoff at: */ for (i=0;i<out_len;i++) { i2=2*i; fstart=i2-(in_len-1); fstart=MAX(fstart,f->start); fend=MIN(i2,f->end); #ifdef TRACE sprintf(dbgstr,"i=%d fstart=%d fend=%d\n",i,fstart,fend); Trace(dbgstr); #endif out_pix=out->data+out_start+i*out_step; in_pix=in->data+in_start+(i2-fstart)*in_step; f_data=f->data-f->start+fstart; for (j=fstart;j<=fend;j++,in_pix-=in_step,f_data++) { *out_pix += (*f_data) * (*in_pix); #ifdef TRACE sprintf(dbgstr," j=%2d in: %4.2f filter: %4.2f [%4d] [%4d] opt : %4.2f %4.2f\n", j, in->data[in_start+in_step*(i2-j)], f->data[j-f->start], in_start+in_step*(i2-j), j-f->start, *in_pix, *f_data); Trace(dbgstr); #endif } } Leaving; return 1; } int filter_inv_cutoff(Image in, int in_start, int in_len, int in_step, Image out, int out_start, int out_len, int out_step, Filter f) { int i,j; Pixel *out_pix, *in_pix, *f_data; int fstart,fend; /* Source interval */ Entering; PreCondition(out_len == in_len*2,"out_len != in_len*2 !!!"); /* convolution: out[i]=sum_{j=start}^{end} (f[2*j-i]*in[j]) boundaries: image in [in_start ... in_start + in_len-1] image out [out_start ... out_start + out_len-1] filter f [0..f->len-1] = [f->start .. f->end] cutoff at: */ for (i=0;i<out_len;i++) { fstart=CEILING_HALF(f->start+i); fend=FLOOR_HALF(f->end+i); fstart=MAX(fstart,0); fend=MIN(fend,in_len-1); #ifdef TRACE sprintf(dbgstr,"i=%d fstart=%d fend=%d\n",i,fstart,fend); Trace(dbgstr); #endif out_pix=out->data+out_start+i*out_step; in_pix=in->data+in_start+fstart*in_step; f_data=f->data-f->start+2*fstart-i; for (j=fstart;j<=fend;j++,in_pix+=in_step,f_data+=2) { *out_pix += (*f_data) * (*in_pix); #ifdef TRACE sprintf(dbgstr," j=%2d in: %4.2f filter: %4.2f [%4d] [%4d] opt : %4.2f %4.2f\n", j, in->data[in_start+j*in_step], f->data[2*j-i-f->start], in_start+j*in_step, 2*j-i-f->start, *in_pix, *f_data); Trace(dbgstr); #endif } } Leaving; return 1; } int filter_periodical(Image in, int in_start, int in_len, int in_step, Image out, int out_start, int out_len, int out_step, Filter f) { int i,i2,j; Pixel *out_pix, *in_pix, *f_data; int fstart,fend; int istart; int ipix_len; Entering; PreCondition(out_len == in_len/2,"out_len != in_len/2 !!!"); /* convolution: out[i]=sum_{j=start}^{end} (in[2*i-j]*f[j]) boundaries: image in [in_start ... in_start + in_len-1] image out [out_start ... out_start + out_len-1] filter f [0..f->len-1] = [f->start .. f->end] */ ipix_len=in_len*in_step; for (i=0;i<out_len;i++) { i2=2*i; fstart=f->start; fend=f->end; istart=(i2-fstart); istart=MOD(istart,in_len); #ifdef TRACE sprintf(dbgstr,"i=%d istart=%d\n",i,istart); Trace(dbgstr); #endif out_pix=out->data+out_start+i*out_step; in_pix=in->data+in_start+istart*in_step; f_data=f->data; for (j=fstart;j<=fend;j++,f_data++) { *out_pix += (*f_data) * (*in_pix); #ifdef TRACE sprintf(dbgstr," j=%2d in: %4.2f filter: %4.2f [%4d] [%4d] opt : %4.2f %4.2f\n", j, in->data[in_start+in_step*((i2-j+in_len)%in_len)], f->data[j-f->start], in_start+in_step*((i2-j+in_len)%in_len), j-f->start, *in_pix, *f_data); Trace(dbgstr); #endif in_pix-=in_step; istart--; if (istart<0) { istart+=in_len; in_pix+=ipix_len; } } } Leaving; return 1; } int filter_inv_periodical(Image in, int in_start, int in_len, int in_step, Image out, int out_start, int out_len, int out_step, Filter f) { int i,j; Pixel *out_pix, *in_pix, *f_data; int fstart,fend; /* Source interval */ int istart; int ipix_len; Entering; PreCondition(out_len == in_len*2,"out_len != in_len*2 !!!"); /* convolution: out[i]=sum_{j=start}^{end} (f[2*j-i]*in[j]) boundaries: image in [in_start ... in_start + in_len-1] image out [out_start ... out_start + out_len-1] filter f [0..f->len-1] = [f->start .. f->end] */ ipix_len=in_len*in_step; for (i=0;i<out_len;i++) { fstart=CEILING_HALF(f->start+i); fend=FLOOR_HALF(f->end+i); istart=MOD(fstart,in_len); #ifdef TRACE sprintf(dbgstr,"i=%d fstart=%d fend=%d istart=%d\n",i,fstart,fend,istart); Trace(dbgstr); #endif out_pix=out->data+out_start+i*out_step; in_pix=in->data+in_start+istart*in_step; f_data=f->data-f->start+2*fstart-i; for (j=fstart;j<=fend;j++,f_data+=2) { *out_pix += (*f_data) * (*in_pix); #ifdef TRACE sprintf(dbgstr," j=%2d in: %4.2f filter: %4.2f [%4d] [%4d] opt : %4.2f %4.2f\n", j, in->data[in_start+(j % in_len)*in_step], f->data[2*j-i-f->start], in_start+(j%in_len)*in_step, 2*j-i-f->start, *in_pix, *f_data); Trace(dbgstr); #endif in_pix+=in_step; istart++; if (istart>=in_len) { istart-=in_len; in_pix-=ipix_len; } } } Leaving; return 1; } int filter_mirror(Image in, int in_start, int in_len, int in_step, Image out, int out_start, int out_len, int out_step, Filter f) { int i,i2,j; Pixel *out_pix, *in_pix, *f_data; int fstart,fend; int in_pos; Entering; PreCondition(out_len == in_len/2,"out_len != in_len/2 !!!"); /* convolution: out[i]=sum_{j=start}^{end} (in[2*i-j]*f[j]) boundaries: image in [in_start ... in_start + in_len-1] image out [out_start ... out_start + out_len-1] filter f [0..f->len-1] = [f->start .. f->end] */ in_pix=in->data+in_start; for (i=0;i<out_len;i++) { i2=2*i; fstart=f->start; fend=f->end; out_pix=out->data+out_start+i*out_step; f_data=f->data; for (j=fstart;j<=fend;j++) { in_pos=(i2-j); if (in_pos<0) { in_pos=-in_pos; if (in_pos>=in_len) continue; } if (in_pos>=in_len) { in_pos=2*in_len-2-in_pos; if (in_pos<0) continue; } *out_pix += (f_data[j-fstart]) * (in_pix[in_pos*in_step]); } } Leaving; return 1; } int filter_inv_mirror(Image in, int in_start, int in_len, int in_step, Image out, int out_start, int out_len, int out_step, Filter f) { int i,j; Pixel *out_pix, *in_pix; int fstart,fend; /* Source interval */ int in_pos; Entering; PreCondition(out_len == in_len*2,"out_len != in_len*2 !!!"); /* convolution: out[i]=sum_{j=start}^{end} (f[2*j-i]*in[j]) boundaries: image in [in_start ... in_start + in_len-1] image out [out_start ... out_start + out_len-1] filter f [0..f->len-1] = [f->start .. f->end] */ /*fprintf(stderr,"inv started\n");*/ for (i=0;i<out_len;i++) { fstart=CEILING_HALF(f->start+i); fend=FLOOR_HALF(f->end+i); out_pix=out->data+out_start+i*out_step; in_pix=in->data+in_start; /* printf("in: %4d - %4d flt: %4d - %4d [%s]\n",fstart,fend,2*fstart-i,2*fend-i, (2*fstart-i<f->start || 2*fend-i>f->end) ? "error":"ok"); */ /*fprintf(stderr,"inv[%d]\n",i);*/ for (j=fstart;j<=fend;j++) { in_pos=j; if (in_pos<0) { if (f->hipass) in_pos=-in_pos-1; else in_pos=-in_pos; if (in_pos>=in_len) continue; } if (in_pos>=in_len) { if (f->hipass) in_pos=2*in_len-2-in_pos; else in_pos=2*in_len-1-in_pos; if (in_pos<0) continue; } /*fprintf(stderr,"out+= %7.2f * %7.2f = %7.2f\n",f->data[2*j-i-f->start],in_pix[in_pos*in_step],f->data[2*j-i-f->start]*in_pix[in_pos*in_step]);*/ *out_pix += f->data[2*j-i-f->start] * (in_pix[in_pos*in_step]); } } Leaving; return 1; } #define MAX_LINE 256 #define skip_blank(str) { while(isspace(*(str))) (str)++; } static int get_next_line(FILE *f, char *c) { char *str,string[200]; int len; do { str=string; if (!fgets(str,200,f)) { Trace("get_next_line: eof\n"); goto error; } len=strlen(str); while (len>=1 && isspace(str[len-1])) str[--len]=0; while (isspace(*str)) str++; } while (strlen(str)==0 || *str=='#'); strcpy(c,str); return 1; error: return 0; } static int next_line_str(FILE *f, char *tag, char *out) { char str[MAX_LINE],*t_str; if (!get_next_line(f,str)) goto error; t_str=strtok(str," "); if (!t_str || strcmp(t_str,tag)) goto error; t_str=strtok(NULL,"\n"); if (!t_str) goto error; skip_blank(t_str); strcpy(out,t_str); return 1; error: return 0; } static int next_line_str_alloc(FILE *f, char *tag, char **out) { char str[MAX_LINE]; if (!next_line_str(f,tag,str)) goto error; *out=malloc(strlen(str)+1); strcpy(*out,str); return 1; error: return 0; } static int next_line_int(FILE *f, char *tag, int *out) { char str[MAX_LINE]; if (next_line_str(f,tag,str) && sscanf(str,"%d",out)==1) return 1; else return 0; } static Filter read_filter(FILE *f) { char str[MAX_LINE]; Filter filter; int i; Entering; filter=calloc(1,sizeof(struct FilterStruct)); if (!next_line_str(f,"Type",str)) goto error1; if (!strcmp(str,"nosymm")) { filter->type=FTNoSymm; } else if (!strcmp(str,"symm")) { filter->type=FTSymm; } else if (!strcmp(str,"antisymm")) { filter->type=FTAntiSymm; } else goto error1; if (!next_line_int(f,"Length",&(filter->len))) goto error1; if (!next_line_int(f,"Start",&(filter->start))) goto error1; if (!next_line_int(f,"End",&(filter->end))) goto error1; if ((filter->end-filter->start+1!=filter->len)) { Trace("error: len != end-start+1\n"); goto error1; } filter->data=calloc(filter->len,sizeof(Pixel)); for (i=0;i<filter->len;i++) { if (!get_next_line(f,str)) goto error2; if (!string_to_pixel(str,filter->data+i)) { Trace("error: invalid filter-value\n"); goto error2; } } if (!get_next_line(f,str)) goto error2; if (*str!='}') { Trace("error: '}' not found\n"); goto error2; } Leaving; return filter; error2: free(filter->data); error1: free(filter); LeavingErr; return NULL; } static FilterGH read_filter_gh(FILE *f) { char str[MAX_LINE]; FilterGH fgh; Filter filter; int i,max; Entering; fgh=calloc(1,sizeof(struct FilterGHStruct)); if (!next_line_str_alloc(f,"Name",&(fgh->name))) { Trace("error: 'Name' tag not found\n"); goto error1; } if (!next_line_str(f,"Type",str)) { Trace("error: 'Type' tag not found\n"); goto error1; } if (!strcmp(str,"orthogonal")) { fgh->type=FTOrtho; max=2; } else if (!strcmp(str,"biorthogonal")) { fgh->type=FTBiOrtho; max=4; } else if (!strcmp(str,"other")) { fgh->type=FTOther; max=4; } else { Trace("error: expecting 'orthogonal', 'biorthogonal' or 'other' type-tag\n"); goto error1; } for (i=0;i<max;i++) { if (!get_next_line(f,str)) goto error2; if (*str!='{') { Trace("error: '{' not found\n"); goto error2; } if (!(filter=read_filter(f))) { Trace("error: read_filter failed\n"); goto error2; } filter->hipass = !(i&1); switch(i) { case 0: fgh->g=filter; break; case 1: fgh->h=filter; break; case 2: fgh->gi=filter; break; case 3: fgh->hi=filter; break; } } if (!get_next_line(f,str)) goto error2; if (*str!='}') { Trace("error: '}' not found\n"); goto error2; } Leaving; return fgh; error2: if (fgh->g) free(fgh->g); if (fgh->h) free(fgh->h); if (fgh->gi) free(fgh->gi); if (fgh->hi) free(fgh->hi); error1: free(fgh); LeavingErr; return NULL; } AllFilters load_filters(char *name) { FILE *f; char str[MAX_LINE]; AllFilters a; FilterGH fgh; Entering; PreCondition(name!=NULL,"name=NULL!"); f=fopen(name,"rt"); if (!f) { Trace("error: fopen failed\n"); goto error1; } a=calloc(1,sizeof(struct AllFilterStruct)); a->count=0; while (get_next_line(f,str)) { if (*str=='{') { fgh=read_filter_gh(f); if (!fgh) { Trace("error: read_filter returned NULL\n"); goto error2; } if (a->count) a->filter=realloc(a->filter,(a->count+1)*sizeof(FilterGH)); else a->filter=malloc(sizeof(FilterGH)); (a->filter)[a->count]=fgh; a->count++; } else { Trace("error: '{' not found\n"); goto error2; } } fclose(f); Leaving; return a; error2: fclose(f); error1: LeavingErr; return 0; } #define doubletree_min 32 #define best_basis_min 8 static int convolute_lines(Image output,Image input,Filter flt,enum FilterMethod method); static int convolute_rows(Image output,Image input,Filter flt,enum FilterMethod method); static int decomposition(Image t_img,Image coarse,Image horizontal,Image vertical, Image diagonal,Filter g,Filter h,enum FilterMethod method); static int compute_best(Image_tree tree,int level,int max_level,FilterGH *flt, enum FilterMethod method,enum Information_Cost cost,double epsilon); static double compute_entropy(Image img,enum Information_Cost cost,double epsilon); static void compute_levels(Image_tree tree,double *entropies,enum Information_Cost cost,double epsilon); static void free_levels(Image_tree tree,int best); static Pixel sumationq(Image img); static Pixel normq(Image_tree tree); static Pixel sumation_down(Image_tree tree, Pixel normq); static Pixel compute_non_additive(Image_tree tree,int size,enum Information_Cost cost,double epsilon,int down); /************************************************************************/ /* Functionname: wavelettransform */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* original: Image that should be transformed */ /* level: transform down to level */ /* flt: transform with filters flt[0..level] */ /* method: method of filtering */ /* -------------------------------------------------------------------- */ /* Description: Carries out the wavelettransform */ /* */ /************************************************************************/ Image_tree wavelettransform(Image original,int level,FilterGH *flt,enum FilterMethod method) { int i,width,height,min,max_level,e; Image coarsei,horizontali,verticali,diagonali,tempi; Image_tree ret_tree,temp_tree; width=original->width; height=original->height; tempi=new_image(width,height); if(!tempi) goto error; copy_into_image(tempi,original,0,0); ret_tree=new_image_tree(); if(!ret_tree) goto error; temp_tree=ret_tree; ret_tree->level=0; min=original->width; if (original->height<min) min=original->height; max_level=log(min)/log(2)-2; if (max_level<level) level=max_level; if (level<1) /* do not transform */ { ret_tree->image=tempi; return ret_tree; } /* decomposition */ for (i=0;i<level;i++) { width=(width+1)/2; height=(height+1)/2; coarsei=new_image(width,height); horizontali=new_image(width,height); verticali=new_image(width,height); diagonali=new_image(width,height); if(!coarsei||!horizontali||!verticali||!diagonali) goto error; e=decomposition(tempi,coarsei,horizontali,verticali,diagonali, flt[i]->g,flt[i]->h,method); if (!e) return NULL; temp_tree->coarse=new_image_tree(); temp_tree->horizontal=new_image_tree(); temp_tree->vertical=new_image_tree(); temp_tree->diagonal=new_image_tree(); temp_tree->coarse->level=i+1; temp_tree->horizontal->level=i+1; temp_tree->vertical->level=i+1; temp_tree->diagonal->level=i+1; temp_tree->horizontal->image=horizontali; temp_tree->vertical->image=verticali; temp_tree->diagonal->image=diagonali; free_image(tempi); if (i!=(level-1)) { tempi=new_image(width,height); copy_into_image(tempi,coarsei,0,0); free_image(coarsei); /*if i=level coarsei is inserted into the image tree so we should not free coarsei on level-1*/ } temp_tree=temp_tree->coarse; } temp_tree->image=coarsei; return ret_tree; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } static Image_tree wavelettransform__wp(Image original, int current_level, int level, FilterGH *flt, enum FilterMethod method) { int i, width, height, min, max_level, e; Image coarse_image,horizontal_image,vertical_image,diagonal_image,temp_image; Image_tree return_tree, temp_tree; width = original->width; height = original->height; temp_image = new_image(width, height); if (!temp_image) goto error; copy_into_image(temp_image, original, 0, 0); temp_tree = return_tree = new_image_tree(); if (!return_tree) goto error; temp_tree->level = current_level; min = original->width; if (original->height < min) min = original->height; max_level = log(min) / log(2) - 2; if (max_level < level) level = max_level; if (current_level >= level) { return_tree->image = temp_image; return return_tree; } for (i = current_level; i < level; i++) { width = (width + 1) / 2; height = (height + 1) / 2; coarse_image = new_image(width, height); horizontal_image = new_image(width, height); vertical_image = new_image(width, height); diagonal_image = new_image(width, height); if (!coarse_image || !horizontal_image || !vertical_image || !diagonal_image) goto error; e = decomposition(temp_image, coarse_image, horizontal_image, vertical_image, diagonal_image, flt[i]->g, flt[i]->h, method); if (!e) return NULL; temp_tree->coarse = new_image_tree(); temp_tree->coarse->level = i+1; temp_tree->horizontal = wavelettransform__wp(horizontal_image, i+1, level, flt, method); temp_tree->vertical = wavelettransform__wp(vertical_image, i+1, level, flt, method); temp_tree->diagonal = wavelettransform__wp(diagonal_image, i+1, level, flt, method); free_image(horizontal_image); free_image(vertical_image); free_image(diagonal_image); free_image(temp_image); if (i != (level - 1)) { temp_image = new_image(width, height); copy_into_image(temp_image, coarse_image, 0, 0); free_image(coarse_image); } temp_tree = temp_tree->coarse; } temp_tree->image = coarse_image; return return_tree; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } Image_tree wavelettransform_wp(Image original, int level, FilterGH *flt, enum FilterMethod method) { return wavelettransform__wp(original, 0, level, flt, method); } /************************************************************************/ /* Functionname: best_basis */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* original: Image to be transformed */ /* level: best basis selection down to this level */ /* flt: transform with filters flt[0..level] */ /* method: transform with filter method */ /* cost: carry best basis selection out with this costfunc */ /* epsilon: limit for threshold method */ /* -------------------------------------------------------------------- */ /* Description: carries best basis and near best basis selection */ /* out */ /************************************************************************/ Image_tree best_basis(Image original,int level,FilterGH *flt, enum FilterMethod method,enum Information_Cost cost,double epsilon) { Image_tree tree; Image img; int min,max_level,e; tree=new_image_tree(); if(!tree) goto error; img=new_image(original->width,original->height); if(!img) goto error; copy_into_image(img,original,0,0); tree->image=img; min=original->width; if (original->height<min) min=original->height; max_level=log10((float) min/best_basis_min)/log10(2); if (max_level>level) max_level=level; e=compute_best(tree,0,max_level,flt,method,cost,epsilon); if (!tree->image) free(img); return tree; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } /************************************************************************/ /* Functionname: best_level_selection */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* original: Image that should be transformed */ /* maxlevel: transform down to level */ /* flt: transform with filters flt[0..level] */ /* method: transform with filter method */ /* cost: carry best basis selection out with this costfunc */ /* epsilon: limit for threshold method */ /* -------------------------------------------------------------------- */ /* Description: Carries out the best level selection */ /* */ /************************************************************************/ Image_tree best_level(Image original,int maxlevel,int *bestlevel,FilterGH *flt,enum FilterMethod method, enum Information_Cost cost,double epsilon) { Image_tree tree; Image img; double *entropies,min; int best=0,i,e; img=new_image(original->width,original->height); copy_into_image(img,original,0,0); tree=new_image_tree(); tree->image=img; entropies=(double *)calloc(maxlevel+1,sizeof(double)); if(!entropies) goto error; /* decompose down to maxlevel */ e=decompose_all(tree,maxlevel,flt,method,cost,epsilon); if (!e) return NULL; /* compute costs of each level and store it in entropies array*/ compute_levels(tree,entropies,cost,epsilon); min=entropies[0]; for (i=1;i<=maxlevel;i++) { if (entropies[i]<min) { min=entropies[i]; best=i; } } /* free all other levels */ free_levels(tree,best); *bestlevel=best; return tree; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } /************************************************************************/ /* Functionname: compute_best */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* img: Image that should be transformed */ /* level: transform level */ /* max_level: transform to maximum level */ /* flt: transform with filters flt[0..level] */ /* method: transform with filter method */ /* cost: carry best basis selection out with this costfunc */ /* epsilon: limit for threshold method */ /* -------------------------------------------------------------------- */ /* Description: Carries out the best basis selection (recursivly) */ /* (is needed by the waveletpacket procedure) */ /************************************************************************/ static int compute_best(Image_tree tree,int level,int max_level,FilterGH *flt, enum FilterMethod method,enum Information_Cost cost,double epsilon) { Image coarse,horizontal,vertical,diagonal; int e,width,height; double sum; tree->level=level; /* non additive cost function*/ if (cost>=shanon) { tree->entropy=compute_non_additive(tree,tree->image->size,cost,epsilon,0); } /*additive cost function*/ else tree->entropy=compute_entropy(tree->image,cost,epsilon); if (level<max_level) { width=(tree->image->width+1)/2; height=(tree->image->height+1)/2; tree->coarse=new_image_tree(); tree->horizontal=new_image_tree(); tree->vertical=new_image_tree(); tree->diagonal=new_image_tree(); coarse=new_image(width,height); horizontal=new_image(width,height); vertical=new_image(width,height); diagonal=new_image(width,height); if(!coarse||!vertical||!horizontal||!diagonal) goto error; e=decomposition(tree->image,coarse,horizontal,vertical,diagonal,flt[0]->g,flt[0]->h,method); if (!e) return 0; tree->coarse->image=coarse; tree->horizontal->image=horizontal; tree->vertical->image=vertical; tree->diagonal->image=diagonal; e=compute_best(tree->coarse,level+1,max_level,flt+1,method,cost,epsilon); e=compute_best(tree->horizontal,level+1,max_level,flt+1,method,cost,epsilon); e=compute_best(tree->vertical,level+1,max_level,flt+1,method,cost,epsilon); e=compute_best(tree->diagonal,level+1,max_level,flt+1,method,cost,epsilon); if (!e) return 0; /*going back in recursion*/ if (cost>=shanon) { sum=compute_non_additive(tree,tree->image->size,cost,epsilon,1); } else sum=(tree->coarse->entropy)+(tree->horizontal->entropy) +(tree->vertical->entropy)+(tree->diagonal->entropy); if (tree->entropy>sum) { tree->entropy=sum; free_image(tree->image); /* take down tree */ tree->image=NULL; } else { /* delete the tree downwards */ free_image_tree(tree->coarse); free_image_tree(tree->horizontal); free_image_tree(tree->vertical); free_image_tree(tree->diagonal); tree->coarse=tree->vertical=tree->horizontal=tree->diagonal=NULL; } } return 1; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } /************************************************************************/ /* Functionname: decompose_all */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: Image tree to be decomposed */ /* maxlevel: decompose down to level */ /* flt: transform with filters flt[0..maxlevel] */ /* method: transform with filter method */ /* cost: cost function for entropy computing */ /* epsilon: limit for threshold method */ /* -------------------------------------------------------------------- */ /* Description: whole decompositing down to maxlevel */ /* The original image must be in tree->image */ /************************************************************************/ int decompose_all(Image_tree tree,int maxlevel,FilterGH *flt,enum FilterMethod method, enum Information_Cost cost,double epsilon) { Image original,coarse,horizontal,vertical,diagonal; int e,width,height,level; if (tree->level<maxlevel) { tree->coarse=new_image_tree(); tree->horizontal=new_image_tree(); tree->vertical=new_image_tree(); tree->diagonal=new_image_tree(); original=tree->image; width=(original->width+1)/2; height=(original->height+1)/2; level=tree->level; coarse=new_image(width,height); horizontal=new_image(width,height); vertical=new_image(width,height); diagonal=new_image(width,height); if(!coarse||!vertical||!horizontal||!diagonal) goto error; e=decomposition(tree->image,coarse,horizontal,vertical,diagonal,flt[0]->g,flt[0]->h,method); if (!e) return 0; tree->coarse->image=coarse; tree->horizontal->image=horizontal; tree->vertical->image=vertical; tree->diagonal->image=diagonal; tree->coarse->entropy=compute_entropy(coarse,cost,epsilon); tree->horizontal->entropy=compute_entropy(horizontal,cost,epsilon); tree->vertical->entropy=compute_entropy(vertical,cost,epsilon); tree->diagonal->entropy=compute_entropy(diagonal,cost,epsilon); tree->coarse->level=tree->horizontal->level= tree->vertical->level=tree->diagonal->level=level+1; e=decompose_all(tree->coarse,maxlevel,flt+1,method,cost,epsilon); e=decompose_all(tree->horizontal,maxlevel,flt+1,method,cost,epsilon); e=decompose_all(tree->vertical,maxlevel,flt+1,method,cost,epsilon); e=decompose_all(tree->diagonal,maxlevel,flt+1,method,cost,epsilon); if (!e) return 0; } return 1; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } /************************************************************************/ /* Functionname: compute_levels */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: Image tree where the entropy should be computed */ /* entropies : array for entropy */ /* cost: carry best basis selection out with this costfunc */ /* epsilon: limit for threshold method */ /* -------------------------------------------------------------------- */ /* Description: Compute the entropies of all decomposition levels */ /************************************************************************/ static void compute_levels(Image_tree tree,double *entropies,enum Information_Cost cost,double epsilon) { if (tree->image){ entropies[tree->level]+=compute_entropy(tree->image,cost,epsilon); } if (tree->coarse) compute_levels(tree->coarse,entropies,cost,epsilon); if (tree->horizontal) compute_levels(tree->horizontal,entropies,cost,epsilon); if (tree->vertical) compute_levels(tree->vertical,entropies,cost,epsilon); if (tree->diagonal) compute_levels(tree->diagonal,entropies,cost,epsilon); } /************************************************************************/ /* Functionname: free_levels */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: Image tree which should be cleaned */ /* best: best level */ /* -------------------------------------------------------------------- */ /* Description: clean the image tree except the best level */ /************************************************************************/ static void free_levels(Image_tree tree,int best) { if (tree->level<best) { free_image(tree->image); tree->image=NULL; free_levels(tree->coarse,best); free_levels(tree->horizontal,best); free_levels(tree->vertical,best); free_levels(tree->diagonal,best); } else { if (tree->coarse) { free_image_tree(tree->coarse); free_image_tree(tree->horizontal); free_image_tree(tree->vertical); free_image_tree(tree->diagonal); tree->coarse=tree->horizontal=tree->vertical=tree->diagonal=NULL; } } } /************************************************************************/ /* Functionname: decompose_to_level */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* original: original image */ /* level: decompose to level */ /* flt: decompos with filters[0..level] */ /* method: transform with filter method */ /* -------------------------------------------------------------------- */ /* Description: Decomposes an image to an certain level and stores */ /* only this level in the returned quadtree */ /************************************************************************/ Image_tree decompose_to_level(Image original,int level,FilterGH *flt,enum FilterMethod method) { Image_tree tree; int e; tree=new_image_tree(); tree->image=original; e=decompose_all(tree,level,flt,method,entropy,1); if (!e) return NULL; free_levels(tree,level); return tree; } /************************************************************************/ /* Functionname: decomposition */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* t_img: Image which should be decomposed */ /* coarse,horizontal,vertical,diagonal: */ /* decomposed images */ /* method: transform with filter method */ /* g,h: the transformation is carried out with these filters*/ /* -------------------------------------------------------------------- */ /* Description: This carries out one wavelettransformation */ /* using waveletfilters. */ /************************************************************************/ static int decomposition(Image t_img,Image coarse,Image horizontal,Image vertical, Image diagonal,Filter g,Filter h,enum FilterMethod method) { Image temp1; /*coarse*/ temp1=new_image(coarse->width,t_img->height); if(!temp1) goto error; convolute_lines(temp1,t_img,h,method); convolute_rows(coarse,temp1,h,method); /*horizontal*/ convolute_rows(horizontal,temp1,g,method); free_image(temp1); /*vertical*/ temp1=new_image(vertical->width,t_img->height); if(!temp1) goto error; convolute_lines(temp1,t_img,g,method); convolute_rows(vertical,temp1,h,method); /*diagonal*/ convolute_rows(diagonal,temp1,g,method); free_image(temp1); return 1; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } /************************************************************************/ /* Functionname: inv_decomposition */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* sum: reconstructed image */ /* coarse,horizontal,vertical,diagonal: images to carry out*/ /* the inverse transformation */ /* flt_gh: transform with filters g and h */ /* method: transform with filter method */ /* -------------------------------------------------------------------- */ /* Description: Carries out the wavelettransform */ /* */ /************************************************************************/ static int inv_decomposition(Image sum,Image coarse,Image horizontal,Image vertical, Image diagonal,FilterGH flt_gh,enum FilterMethod method) { Image temp1; Filter g,h; if (flt_gh->type==FTOrtho) { g=flt_gh->g; h=flt_gh->h; } else { g=flt_gh->gi; h=flt_gh->hi; } /*coarse*/ temp1=new_image(coarse->width,sum->height); if(!temp1) goto error; convolute_rows(temp1,coarse,h,method); /*horizontal*/ convolute_rows(temp1,horizontal,g,method); convolute_lines(sum,temp1,h,method); free_image(temp1); /*vertical*/ temp1=new_image(vertical->width,sum->height); if(!temp1) goto error; convolute_rows(temp1,vertical,h,method); /*diagonal*/ convolute_rows(temp1,diagonal,g,method); convolute_lines(sum,temp1,g,method); free_image(temp1); return 1; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } /************************************************************************/ /* Functionname: build_image */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* quadtree: quadtree with decomposition information */ /* width,height: image width and height */ /* RETURN: returns the build up image */ /* -------------------------------------------------------------------- */ /* Description: builds up an image out of an Image_tree */ /* */ /************************************************************************/ Image build_image(Image_tree quadtree,int width,int height) { Image ret_img,coarse,horizontal,vertical,diagonal; ret_img=new_image(width,height); if(!ret_img) goto error; width=(width+1)/2; height=(height+1)/2; if (!(quadtree->image)) { coarse=build_image(quadtree->coarse,width,height); horizontal=build_image(quadtree->horizontal,width,height); vertical=build_image(quadtree->vertical,width,height); diagonal=build_image(quadtree->diagonal,width,height); if (!coarse||!horizontal||!vertical||!diagonal) return NULL; copy_into_image(ret_img,coarse,0,0); copy_into_image(ret_img,horizontal,width,0); copy_into_image(ret_img,vertical,0,height); copy_into_image(ret_img,diagonal,width,height); if (!quadtree->coarse->image) free_image(coarse); if (!quadtree->horizontal->image) free_image(horizontal); if (!quadtree->vertical->image) free_image(vertical); if (!quadtree->diagonal->image) free_image(diagonal); return ret_img; } else return quadtree->image; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } /************************************************************************/ /* Functionname: inv_transform */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: tree with decomposition information */ /* flt_gh: transform with filters g and h */ /* method: transform with filter method */ /* -------------------------------------------------------------------- */ /* Description: Inverts the wavelettransform,best_basis,best_level */ /* */ /************************************************************************/ Image inv_transform(Image_tree tree,FilterGH *flt, enum FilterMethod method) { int er,width,height; Image ret_img,coarse,vertical,horizontal,diagonal; if (!tree->image) { coarse=inv_transform(tree->coarse,flt,method); horizontal=inv_transform(tree->horizontal,flt,method); vertical=inv_transform(tree->vertical,flt,method); diagonal=inv_transform(tree->diagonal,flt,method); if (!coarse||!horizontal||!vertical||!diagonal) return NULL; width=coarse->width+horizontal->width; height=coarse->height+vertical->height; ret_img=new_image(width,height); if(!ret_img) goto error; if (tree->flag==0) /*if flag is set it is a doubletree tiling*/ { // er=inv_decomposition(ret_img,coarse,horizontal,vertical,diagonal,flt[1],method); er=inv_decomposition(ret_img,coarse,horizontal,vertical,diagonal,flt[tree->level],method); if (!er) return NULL; } else { copy_into_image(ret_img,coarse,0,0); copy_into_image(ret_img,horizontal,coarse->width,0); copy_into_image(ret_img,vertical,0,coarse->height); copy_into_image(ret_img,diagonal,coarse->width,coarse->height); } if (!tree->coarse->image) free_image(coarse); if (!tree->horizontal->image) free_image(horizontal); if (!tree->vertical->image) free_image(vertical); if (!tree->diagonal->image) free_image(diagonal); return ret_img; } else return tree->image; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return NULL; } /************************************************************************/ /* Functionname: find_deepest_level */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* -------------------------------------------------------------------- */ /* Description: Finds the deepest possible level where width and */ /* height can divided by two exactly. */ /************************************************************************/ int find_deepest_level(int width,int height) { int level=0,w=width,h=height; while ( !((w%2)||(h%2))) { w=w/2; h=h/2; level++; } return level-1; } /************************************************************************/ /* Functionname: convolute_lines */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* output: output image of wavelettransformation */ /* input: input image for decomposition */ /* flt: transform with filter flt */ /* method: transform with filter method */ /* -------------------------------------------------------------------- */ /* Description: Carries out the wavelettransform for all lines of */ /* the input image */ /************************************************************************/ static int convolute_lines(Image output,Image input,Filter flt,enum FilterMethod method) /*Convolute the lines with filter*/ { int i; for (i=0;i<input->height;i++) { switch(method) { case cutoff: filter_cutoff(input,input->width*i,input->width,1, output,output->width*i,output->width,1,flt); break; case inv_cutoff: filter_inv_cutoff(input,input->width*i,input->width,1, output,output->width*i,output->width,1,flt); break; case periodical: filter_periodical(input,input->width*i,input->width,1, output,output->width*i,output->width,1,flt); break; case inv_periodical: filter_inv_periodical(input,input->width*i,input->width,1, output,output->width*i,output->width,1,flt); break; case mirror: filter_mirror(input,input->width*i,input->width,1, output,output->width*i,output->width,1,flt); break; case inv_mirror: filter_inv_mirror(input,input->width*i,input->width,1, output,output->width*i,output->width,1,flt); break; } } return 1; } /************************************************************************/ /* Functionname: convolute_rows */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* output: output image of wavelettransformation */ /* input: input image for decomposition */ /* flt: transform with filter flt */ /* method: transform with filter method */ /* -------------------------------------------------------------------- */ /* Description: Carries out the wavelettransform for all rows of */ /* the input image */ /************************************************************************/ static int convolute_rows(Image output,Image input,Filter flt,enum FilterMethod method) /*Convolute the rows with filter*/ { int i; for (i=0;i<input->width;i++) { switch (method) { case cutoff: filter_cutoff(input,i,input->height,input->width, output,i,output->height,output->width,flt); break; case inv_cutoff: filter_inv_cutoff(input,i,input->height,input->width, output,i,output->height,output->width,flt); break; case periodical: filter_periodical(input,i,input->height,input->width, output,i,output->height,output->width,flt); break; case inv_periodical: filter_inv_periodical(input,i,input->height,input->width, output,i,output->height,output->width,flt); break; case mirror: filter_mirror(input,i,input->height,input->width, output,i,output->height,output->width,flt); break; case inv_mirror: filter_inv_mirror(input,i,input->height,input->width, output,i,output->height,output->width,flt); break; } } return 1; } /************************************************************************/ /* Functionname: sumationq */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* img: image to compute */ /* -------------------------------------------------------------------- */ /* Description: compute the sum of quadrats of all elements of */ /* the input image */ /************************************************************************/ static Pixel sumationq(Image img) { Pixel sum=0; int i; for (i=0;i<img->size;i++) { sum+=(*img->data+i)*(*img->data+i); } return sum; } /************************************************************************/ /* Functionname: normq */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: tree to compute */ /* -------------------------------------------------------------------- */ /* Description: computes the quadratic norm over all images in */ /* the input tree */ /************************************************************************/ static Pixel normq(Image_tree tree) { Pixel sum=0; if (tree->image) { sum=sumationq(tree->image); } else { if (tree->coarse) sum+=normq(tree->coarse); if (tree->horizontal) sum+=normq(tree->horizontal); if (tree->vertical) sum+=normq(tree->vertical); if (tree->diagonal) sum+=normq(tree->diagonal); } return sum; } /************************************************************************/ /* Functionname: sumation_down */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: tree to compute */ /* normq: norm of the images in the tree */ /* -------------------------------------------------------------------- */ /* Description: computes the Entropy over all (string aded) images */ /* in the input tree */ /************************************************************************/ static Pixel sumation_down(Image_tree tree, Pixel normq) { Pixel sum=0,p; int i; Image img; Pixel *data; if (tree->image) { img=tree->image; data=img->data; for (i=0;i<img->size;i++,data++) { if (*data!=0) { p=(*data)*(*data)/normq; sum+=p*log(1/p); } } } else { if (tree->coarse) sum+=sumation_down(tree->coarse,normq); if (tree->horizontal) sum+=sumation_down(tree->horizontal,normq); if (tree->vertical) sum+=sumation_down(tree->vertical,normq); if (tree->diagonal) sum+=sumation_down(tree->diagonal,normq); } return sum; } /************************************************************************/ /* Functionname: comp */ /* -------------------------------------------------------------------- */ /* Description: used for quicksort for decreasing order */ /************************************************************************/ int comp(const Pixel *x,const Pixel *y) { if (*x<*y) return 1; else if (*x==*y) return 0; else return -1; } /************************************************************************/ /* Functionname: recarea */ /* tree: Image tree to compute */ /* list: target list */ /* list_size: actual size of the list */ /* -------------------------------------------------------------------- */ /* Description: copies all elements within the tree into an list */ /************************************************************************/ static void recarea(Image_tree tree,Pixel *list,int *list_size) { Image img; if (tree->image) { img=tree->image; memcpy(list+(*list_size),img->data,img->size*sizeof(Pixel)); *list_size+=img->size; } else { if (tree->coarse) recarea(tree->coarse,list,list_size); if (tree->horizontal) recarea(tree->horizontal,list,list_size); if (tree->vertical) recarea(tree->vertical,list,list_size); if (tree->diagonal) recarea(tree->diagonal,list,list_size); } } static void abs_list(Pixel *list,int list_size) { int i; for (i=0;i<list_size;i++) list[i]=fabs(list[i]); } /************************************************************************/ /* Functionname: sum_list */ /* -------------------------------------------------------------------- */ /* Description: computes the sum of all poweres list elements */ /************************************************************************/ static Pixel sum_list(Pixel *list,int p,int size) { Pixel sum=0; int i; for (i=0;i<size;i++) { if (p!=1) sum+=pow(list[i],p); else sum+=list[i]; } return sum; } static Pixel weak_lp(Image_tree tree,int size,int p,double epsilon) { Pixel wlp,*list,max=0; int *list_size,k; list_size=(int *)malloc(sizeof(int)); if (!list_size) goto error; *list_size=0; list=(Pixel *)calloc(size,sizeof(Pixel)); if (!list) goto error; recarea(tree,list,list_size); abs_list(list,*list_size); qsort(list,*list_size, sizeof(Pixel), (int (*)(const void*, const void*)) comp); for (k=0;k<size;k++) { if (k!=0) wlp=pow(k,1/p)*list[k]; else wlp=0; if (wlp>max) max=wlp; } free(list); return max; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } static Pixel comp_number(Image_tree tree,int size,int p,double f) { Pixel sum=0,*list,min=MAXDOUBLE,norm,npf,normf; int *list_size=0,k; list_size=(int *)malloc(sizeof(int)); if (!list_size) goto error; *list_size=0; list=(Pixel *)calloc(size,sizeof(Pixel)); if (!list) goto error; recarea(tree,list,list_size); abs_list(list,*list_size); qsort(list,*list_size, sizeof(Pixel), (int (*)(const void*, const void*)) comp); norm=sum_list(list,p,size); normf=norm*f; for (k=0;k<size;k++) { if (list[k]!=0) { sum+=pow(list[k],p); npf=fabs(sum-normf); if (npf<min) min=npf; } } min=min/norm; free(list); return min; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } static Pixel comp_area(Image_tree tree,int size,int p,double f) { Pixel sum=0,*list,norm,vk=0; int *list_size=0,k; list_size=(int *)malloc(sizeof(int)); if (!list_size) goto error; *list_size=0; list=(Pixel *)calloc(size,sizeof(Pixel)); if (!list) goto error; recarea(tree,list,list_size); abs_list(list,*list_size); qsort(list,*list_size, sizeof(Pixel), (int (*)(const void*, const void*)) comp); norm=sum_list(list,p,size); for (k=0;k<size;k++) { if (list[k]!=0) { vk+=pow(list[k],p); sum+=vk; } } free(list); return (size-sum/norm); error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } static Pixel compute_sdiscrepancy(Image_tree tree,int size) { Pixel *list,min,max,factor,maximum=0,x; int *list_size=0,k; list_size=(int *)malloc(sizeof(int)); if (!list_size) goto error; *list_size=0; list=(Pixel *)calloc(size,sizeof(Pixel)); if (!list) goto error; recarea(tree,list,list_size); qsort(list,*list_size, sizeof(Pixel), (int (*)(const void*, const void*)) comp); min=list[0]; max=list[size-1]; factor=1/(max-min); /*scaling to [0,1]*/ for (k=0;k<size;k++) { list[k]=factor*(list[k]-min); } for (k=0;k<size;k++) { x=fabs(list[k]-(2*k-1)/(2*size)); if (x>maximum) maximum=x; } free(list); return (1/(2*size)+maximum); error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } static Pixel compute_discrepancy(Image_tree tree,int size) { Pixel *list,min,max,factor,maximum=0,minimum=0,x; int *list_size=0,k; list_size=(int *)malloc(sizeof(int)); if (!list_size) goto error; *list_size=0; list=(Pixel *)calloc(size,sizeof(Pixel)); if (!list) goto error; recarea(tree,list,list_size); qsort(list,*list_size, sizeof(Pixel), (int (*)(const void*, const void*)) comp); min=list[0]; max=list[size-1]; factor=1/(max-min); /*scaling to [0,1]*/ for (k=0;k<size;k++) { list[k]=factor*(list[k]-min); } for (k=0;k<size;k++) { x=((Pixel)k/size-list[k]); if (x>maximum) maximum=x; else if (x<minimum) minimum=x; } free(list); return (1/size+maximum-minimum); error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } static Pixel compute_concentration(Image_tree tree,int size) { Pixel *list,min,max,factor,lkm=0,length,sum=0,value,norm; int *list_size=0,k,next=0,last=0,i=0; list_size=(int *)malloc(sizeof(int)); if (!list_size) goto error; *list_size=0; list=(Pixel *)calloc(size+1,sizeof(Pixel)); if (!list) goto error; recarea(tree,list,list_size); qsort(list,*list_size, sizeof(Pixel), (int (*)(const void*, const void*)) comp); min=list[0]; max=list[size-1]; length=(max-min)/100; factor=1/(max-min); for (k=0;k<size;k++) { list[k]=factor*(list[k]-min); } norm=size*sum_list(list,1,size); length=0.01; value=length; list[size]=max+value; for (k=0;k<100;k++) { while ((list[i]<value)&(i<size)) { sum+=list[i]; next++; i++; } lkm+=(next-last)*sum/norm; value+=length; last=next; sum=0; } return -lkm; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } /************************************************************************/ /* Functionname: compute_entropy */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* img: Image from which the entropy should be computed */ /* cost: choosen costfunction */ /* epsilon: limit for threshold method */ /* -------------------------------------------------------------------- */ /* Description: computes entropy of an image */ /************************************************************************/ static double compute_entropy(Image img,enum Information_Cost cost,double epsilon) { double sum=0,x=0; int i; Pixel *data; data=img->data; switch(cost) { case threshold: for(i=0;i<img->size;i++) if (fabs(img->data[i])>epsilon) sum++; break; case log_energy: for(i=0;i<img->size;i++,data++) { x=(*data) * (*data); if (x!=0) sum+=(x*log(1/x)); } break; case norml: for(i=0;i<img->size;i++,data++) { x=fabs(*data); sum+=x; } break; case norml2: for(i=0;i<img->size;i++,data++) { x=(*data) * (*data); sum+=x; } sum=pow(sum,0.5); break; case entropy: for(i=0;i<img->size;i++,data++) { x=(*data)*(*data); if (x!=0) sum-=(x*log(x)); } break; case gauss_markov: for(i=0;i<img->size;i++) { x=(img->data[i])*(img->data[i]); if (x!=0) sum+=log(x*x); } break; } return sum; } /************************************************************************/ /* Functionname: compute_non_additive */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* tree: Image tree from which the entropy should be */ /* computed */ /* size : size of the image */ /* cost: choosen costfunction */ /* epsilon: limit for threshold method */ /* down: decides if only the first image should be computed*/ /* -------------------------------------------------------------------- */ /* Description: computes entropy of an image */ /************************************************************************/ static Pixel compute_non_additive(Image_tree tree,int size,enum Information_Cost cost,double epsilon,int down) { Pixel sum=0,normx; Image img=NULL; if (down) { img=tree->image; tree->image=NULL; } switch (cost) { case shanon: normx=normq(tree); sum=-sumation_down(tree,normx); break; case weak_l: sum=weak_lp(tree,size,1,epsilon); break; case weak_lq: sum=weak_lp(tree,size,2,epsilon); break; case compression_number: sum=comp_number(tree,size,1,epsilon); break; case compression_numberq: sum=comp_number(tree,size,2,epsilon); break; case compression_area: sum=comp_area(tree,size,1,epsilon); break; case compression_areaq: sum=comp_area(tree,size,2,epsilon); break; case discrepancy: sum=compute_discrepancy(tree,size); break; case sdiscrepancy: sum=compute_sdiscrepancy(tree,size); break; case concentration: sum=compute_concentration(tree,size); break; } if (down) tree->image=img; return sum; } int rec_double(Image_tree dtree,int level,FilterGH *flt,enum FilterMethod method,enum Information_Cost cost,double epsilon) { int min,width,height; double sum=0; Image c,h,v,d; dtree->level=0; if (cost>=shanon) { dtree->entropy=compute_non_additive(dtree,dtree->image->size,cost,epsilon,0); } else dtree->entropy=compute_entropy(dtree->image,cost,epsilon); dtree->doubletree=best_basis(dtree->image,level,flt,method,cost,epsilon); min=dtree->image->width; if (dtree->image->height<min) min=dtree->image->height; if (doubletree_min<min) { width=(dtree->image->width+1)/2; height=(dtree->image->height+1)/2; dtree->coarse=new_image_tree(); dtree->horizontal=new_image_tree(); dtree->vertical=new_image_tree(); dtree->diagonal=new_image_tree(); c=new_image(width,height); h=new_image(width,height); v=new_image(width,height); d=new_image(width,height); if(!c||!h||!v||!d) goto error; copy_part_of_image(c,dtree->image,0,0); copy_part_of_image(h,dtree->image,width,0); copy_part_of_image(v,dtree->image,0,height); copy_part_of_image(d,dtree->image,width,height); dtree->coarse->image=c; dtree->horizontal->image=h; dtree->vertical->image=v; dtree->diagonal->image=d; rec_double(dtree->coarse,level,flt,method,cost,epsilon); rec_double(dtree->horizontal,level,flt,method,cost,epsilon); rec_double(dtree->vertical,level,flt,method,cost,epsilon); rec_double(dtree->diagonal,level,flt,method,cost,epsilon); /* going back in recursion*/ sum=dtree->coarse->entropy+dtree->horizontal->entropy+ dtree->vertical->entropy+dtree->diagonal->entropy; if (sum>dtree->entropy) { /*take image*/ free_image_tree(dtree->coarse); free_image_tree(dtree->horizontal); free_image_tree(dtree->vertical); free_image_tree(dtree->diagonal); dtree->coarse=dtree->horizontal=dtree->vertical=dtree->diagonal=NULL; } else { /*take tiling*/ dtree->entropy=sum; free_image(dtree->image); dtree->image=NULL; } if (dtree->entropy>dtree->doubletree->entropy) { /*take best basis tree*/ dtree->entropy=dtree->doubletree->entropy; if(dtree->coarse) free_image_tree(dtree->coarse); if(dtree->horizontal) free_image_tree(dtree->horizontal); if(dtree->vertical) free_image_tree(dtree->vertical); if(dtree->diagonal) free_image_tree(dtree->diagonal); dtree->coarse=dtree->doubletree->coarse; dtree->horizontal=dtree->doubletree->horizontal; dtree->vertical=dtree->doubletree->vertical; dtree->diagonal=dtree->doubletree->diagonal; free_image(dtree->image); dtree->image=NULL; free(dtree->doubletree); dtree->doubletree=NULL; } else { dtree->flag=1; if(dtree->doubletree) free_image_tree(dtree->doubletree); dtree->doubletree=NULL; } } return 1; error: err_SimpleMessage(err_GetErrorMessage(Error_NotEnoughMemory)); return 0; } static void save_structur(Image_tree tree,FILE *fp,int pos) { int shift,next_pos,max; if (tree->flag) { fprintf(fp,"%d ",pos); shift=pos-(pow(4,tree->level-1)-1)*4/3-1; max=(int) ((pow(4,tree->level)-1)*4/3); next_pos=max+4*shift+1; if (tree->coarse) save_structur(tree->coarse,fp,next_pos); if (tree->horizontal) save_structur(tree->horizontal,fp,next_pos+1); if (tree->vertical) save_structur(tree->vertical,fp,next_pos+2); if (tree->diagonal) save_structur(tree->diagonal,fp,next_pos+3); } } static int is_in_list(int *list,int len, int x) { int i,found=0; for (i=0;i<len;i++) { if (list[i]==x) { found=1; i=len; } } return found; } static void write_flags(Image_tree tree,int *list,int len,int pos) { int shift,next_pos,max; if (is_in_list(list,len,pos)) { tree->flag=1; shift=pos-(pow(4,tree->level-1)-1)*4/3-1; max=(int) ((pow(4,tree->level)-1)*4/3); next_pos=max+4*shift+1; write_flags(tree->coarse,list,len,next_pos); write_flags(tree->horizontal,list,len,next_pos+1); write_flags(tree->vertical,list,len,next_pos+2); write_flags(tree->diagonal,list,len,next_pos+3); } } /************************************************************************/ /* Functionname: err_simple_message */ /* -------------------------------------------------------------------- */ /* Parameter: */ /* char *: string that contains information about an */ /* error the user should know. */ /* -------------------------------------------------------------------- */ /* Description: */ /* Prints error messages for the user. */ /************************************************************************/ void err_SimpleMessage(char *message) { fprintf(stderr,"%s\n",message); } /************************************************************************/ /* Functionname: err_get_message */ /* -------------------------------------------------------------------- */ /* Return value: Errormessage for this specific error. */ /* Parameter: */ /* Error err: Error whose errormessage should be returned */ /* -------------------------------------------------------------------- */ /* Description: */ /************************************************************************/ char * err_GetErrorMessage(Error err) { switch (err) { case Error_NotImplemented: return "Sorry, this is not implemented yet. "; break; case Error_AssertionFailed: return "Sorry, an internal assertion was violated.\n" "This action can not be completed. :-("; break; case Error_NotEnoughMemory: return "Sorry, there is not enough memory"; break; case Error_Limitation: return "Some limitation of the program exceeded"; break; /* - FILES - */ case Error_CantOpenFile: return "Could not open file"; break; case Error_CantCreateFile: return "Could not create file"; break; case Error_CantCloseFile: return "Could not close file"; break; case Error_InternalError: return "Sorry, an internal error occured.\n" "This action can not be completed. :-("; break; default: return "Sorry, but an unknown error ocurred.\n" "This action can not be completed. :-("; break; } }