Mercurial > hg > wm
comparison Fotopoulos/README_VASSILIS @ 0:be303a3f5ea8
import
| author | Peter Meerwald <pmeerw@cosy.sbg.ac.at> |
|---|---|
| date | Sun, 12 Aug 2007 13:14:34 +0200 |
| parents | |
| children |
comparison
equal
deleted
inserted
replaced
| -1:000000000000 | 0:be303a3f5ea8 |
|---|---|
| 1 These codes use the DCT,Hartley and Subband DCT | |
| 2 Transforms for watermarking purposes.All schemes | |
| 3 are blind,no original image is used for detection. | |
| 4 Each casting module is accompanied by a testing | |
| 5 module.Supposing that the watermark key that you | |
| 6 select during casting is in the range 1-1000,the | |
| 7 testing module will test all the keys in this range | |
| 8 to produce the well known similarity diagrams used | |
| 9 by Cox, Piva and many more with the peak being the | |
| 10 proof of the watermark existence. | |
| 11 | |
| 12 I do not include a detection module because this implies | |
| 13 the use of certain thresholds. Although the casting | |
| 14 methods are almost standardized (multiplicative formula) | |
| 15 there are still questions about this thresholding but | |
| 16 the testing modules can be easily changed to fit this | |
| 17 purpose. | |
| 18 | |
| 19 The DCT scheme does not use the visual masking improvement | |
| 20 that Barni and his team suggest in one of their later works. | |
| 21 In all casting programs we assume that the coefficients are | |
| 22 diagonaly scanned,and ordered as shown in the following example. | |
| 23 | |
| 24 ---------------------- | |
| 25 | 1| 3| 6|10|15|... | |
| 26 ---------------------- | |
| 27 | 2| 5| 9|14|... | |
| 28 ---------------------- | |
| 29 | 4| 8|13|... | |
| 30 ---------------------- | |
| 31 | 7|12|... | |
| 32 ---------------------- | |
| 33 |11|... | |
| 34 ---------------------- | |
| 35 | |
| 36 which is quite simple comparing to the zig zag scanning | |
| 37 pattern that we know from the JPEG standard but does not | |
| 38 affect at all the idea that we have of the middle | |
| 39 frequencies.Also the user should take care that | |
| 40 | |
| 41 starting_coefficient+number_of_coeffs_to_change<(N*N)/2 | |
| 42 | |
| 43 which means that we shouldn't exceed the matrix diagonal. | |
| 44 To do so the scanning scheme should be changed accordingly | |
| 45 but this doesn't seem important because in all of the | |
| 46 schemes we don't get out of the middle frequencies coefs. | |
| 47 With correctly selected parameters,the schemes perform in | |
| 48 the same way as if the coeffs were zig zag scanned. | |
| 49 | |
| 50 In the Subband DCT version,the first set of questions | |
| 51 about starting coefficient,number of coefficients to | |
| 52 alter and alpha parameter, refer to the LH,HL and HH | |
| 53 band while the second set of questions sets the parameters | |
| 54 about the LL band only. | |
| 55 | |
| 56 All schemes should not be used with the same set of parameters | |
| 57 because each of the transforms, possess certain specific | |
| 58 properties. This should be kept in mind for testing purposes. | |
| 59 To use the right set of parameters please refer to corresponding | |
| 60 bibliography. | |
| 61 All code has been tested in Visual C++ v6.0 | |
| 62 | |
| 63 Have some nice tests... | |
| 64 Vassilis Fotopoulos | |
| 65 | |
| 66 for more info,ideas or points of discussion | |
| 67 email vfotop1@physics.upatras.gr |