Presentation is loading. Please wait.

Presentation is loading. Please wait.

Bit-Plane Watermarking for SPIHT-Coded Images 台北科技大學資工所 指導教授:楊士萱 學生:廖武傑 2003/07/29.

Similar presentations


Presentation on theme: "Bit-Plane Watermarking for SPIHT-Coded Images 台北科技大學資工所 指導教授:楊士萱 學生:廖武傑 2003/07/29."— Presentation transcript:

1 Bit-Plane Watermarking for SPIHT-Coded Images 台北科技大學資工所 指導教授:楊士萱 學生:廖武傑 2003/07/29

2 OUTLINE Introduction to watermarking SPIHT Factors that affect watermarking system Proposed watermarking scheme Conclusion

3 Watermarking The watermark is an imperceptible but indelible code used for ownership identification. Why? convenience of digital multimedia data intellectual property of digitally recorded material shortcomings of encryption techniques Requirements perceptual transparency robustness security appropriate complexity Computational requirements introduced by watermark embedding and detection should be small

4 Domains of image watermarking

5 OUTLINE Introduction to watermarking SPIHT Factors that affect watermarking system Proposed watermarking scheme Conclusion

6 SPIHT coding procedure

7 Wavelet transform(1D) h0 h1 h0 h1 2 2 2 2 Resolution 1/2 Resolution 1/4 C 3,n C 2,n C 1,n d 2,n d 1,n

8 Wavelet transform(2D) h0 h1 2 2 2 2 2 2 Horizontal filteringVertical filtering

9 Wavelet filter Daubechies 9/7 Analysis Filter Coefficients: iLow-Pass FilterHigh-Pass Filter 00.60294901823635791.115087052456994 ±10.2668641184428723-0.5912717631142470 ±2-0.078223266652898785-0.05754352622849957 ±3-0.016864118442874950.09127176311424948 ±40.02674875741080976

10 Example, “Lena”

11 SPIHT coding procedure

12 Spatial Orientation Trees

13 Example (1 st Sorting pass, Encoder) LIPLSP LIS: T=32 A2A3A4 1+1-0011+000

14 Example (1 st Sorting pass) LIPLSP LIS: T=32 A3A4 1+1-0011+000 B2 10000

15 Example (1 st Sorting pass) LIPLSP LIS: T=32 A4 1+1-0011+000 B2 10000 B3 001

16 Example (1 st Sorting pass) LIPLSP LIS: T=32 A4 1+1-0011+000 B2 10000 A9 001 A10A11A12 001+001

17 Example (1 st Sorting pass) LIPLSP LIS: T=32 A4 1+1-0011+000 B2 10000 A9 001 A11A12 001+00001

18 Example (1 st Refinement pass) Do nothing

19 Example (2nd Sorting pass) LIPLSP LIS: T=16 A4B2A9A11A12 1-1+000000000000000

20 Example (2nd Refinement pass) LSP: In 1st sorting pass: 63 +111111 -34 -100010 49 +110001 47 +101111 -31 - 11111 23 + 10111 1-1+0000000000000001010

21 Example (3rd Sorting pass) LIPLSPT=8

22 Example (4th Sorting pass) LIPLSPT=4

23 Example (5th Sorting pass) LIPLSPT=2

24 Example (1 st Sorting pass, Decoder) Received: 1+1-0011+00010000001001+00001 originaldecoded

25 Example (1 st Sorting pass, Decoder) 63 64 32 1 Original coefficients Reconstructed value (by 1st sorting pass) 48

26 Example (2 nd Sorting pass, Decoder) Received: originaldecoded 1-1+000000000000000

27 Example (2 nd Refinement pass, Decoder) 63 64 32 1 Original coefficients Reconstructed value (by 1st sorting pass) 48 10 Refinement bit: 56 Refinement by 2nd refinement pass 40 If refinement bit is ‘0’

28 Example (2 nd Refinement pass, Decoder) Received: originaldecoded 1010

29 OUTLINE Introduction to watermarking SPIHT Factors that affect watermarking system Proposed watermarking scheme Conclusion

30 Watermarks over a noise channel The noise comes from:  The host signal  Lossy compression  Other attacks

31 Host-interference-nonrejecting watermarking method Transform domain watermarking Watermark: Gaussian random source Embedding method: After distortion: Effect of host interference and quantization noise W. Zhu, Z. Xiong, and Y.-Q. Zhang, “Mutiresolution watermarking for images and video,” IEEE Trans. Circuits Syst. Video Technol., June 1999

32 Watermark similarity measure for judging: (linear correlation) where: Informed detection: Effect of host interference and quantization noise (cont.)

33 Test images “Lena”“Baboon”

34 Degradation of watermark similarity due to quantization Bpp1/641/321/161/81/41/21 Step size5122561286432168 MSE PSNR 357.46 22.60 205.63 25.00 122.29 27.26 66.66 29.89 37.73 32.36 23.12 34.49 15.91 36.12 Correlati on (informe d) 0.71690.78240.91230.97580.99390.99840.9996 Correlati on (blind) 0.4444 43997.8 0.3189 30494.1 0.3381 307690.2 0.3358 30402.3 0.3373 30580.6 0.3385 30660.7 0.3382 30641.3 “Lena”, alpha=0.26, watermark length:255, subband:1-6, coefficients smaller than 512

35 Degradation of watermark similarity due to quantization “Baboon”, alpha=0.28, watermark length:255, subband:1-6, coefficients smaller than 512 Bpp1/641/321/161/81/41/21 Step size512256 128 6432 MSE PSNR 757.05 19.34 662.30 19.92 582.58 20.48 468.47 21.42 343.12 22.78 209.20 24.93 98.03 28.22 Correlati on (infor med) 0.71060.73510.76290.88390.88410.96510.9917 Correlati on (blin d) 0.3917 32402.5 0.2267 18845 0.2130 17572 0.2088 16264.5 0.2089 16274.5 0.2229 17283.9 0.2219 17182.5

36 OUTLINE Introduction to watermarking SPIHT Factors that affect watermarking system Proposed watermarking scheme Conclusion

37 Proposed watermarking scheme

38 Proposed watermarking scheme (Cont.) Embedding our watermark

39 The effect of wavelet coefficient after embedding watermark

40 Extracting watermark procedure

41 Watermarking detecting

42 Proposed watermarking scheme (Cont.) Host-interference rejection Minimization of quantization effects Low complexity Progressive nature

43 StirMark attacks JPEG compression Gaussian filtering Median filtering Sharpening FMLR

44 JPEG compression attacks

45 Gaussian filtering

46 Median filtering

47 Sharpening

48 FMLR

49 watermark length:255 PN code, subband:1-6, coefficient between 512 and 256 soft-decision, bit-rate: 0.25bpp LenaBaboonPepperF16 Jpeg 50% PSNR 0.9994/0.9925 /0.3300 31.78 0.9994/0.8821 /0.2094 22.48 0.9971/0.9817 0.2477 30.45 0.9996/0.9936 /0.2937 31.00 Jpeg 30% PSNR 0.9986/0.9914 /0.3305 31.34 0.9987/0.8783 /0.2089 22.32 0.9923/0.9796 /0.2476 30.09 0.9989/0.9917 /0.2941 30.63 Jpeg 10% PSNR 0.9891/0.9619 /0.3182 29.27 0.9856/0.8538 /0.2064 21.58 0.9784/0.9797 /0.2442 28.48 0.9897/0.9720 /0.2865 28.71 Gaussian filter PSNR 0.9987/0.9887 /0.3303 31.05 0.9973/0.8831 /0.2072 22.26 0.9915/0.9719 /0.2459 29.48 0.9970/0.9857 /0.2931 29.58 Median filter PSNR 0.8747/0.7519 /0.3291 25.42 0.9835/0.8536 /0.2038 21.79 0.9546/0.9093 /0.2471 27.64 0.9785/0.9455 /0.2839 27.80 Sharpening PSNR 0.9876/0.9437 /0.3293 23.99 0.9723/0.8040 /0.2060 17.24 0.9713/0.9143 /0.2545 24.49 0.9852/0.9288 /0.2912 23.94 FMLR PSNR 0.9373/0.7917 /0.3202 29.76 0.9781/0.8097 /0.2084 22.64 0.9124/0.7522 /02109 28.27 0.9626/0.8741 /0.2792 29.65

50 watermark length:255 PN code, subband:1-6, coefficient between 512 and 256, soft-decision, bit-rate: 0.5bpp LenaBaboonPepperF16 Jpeg 50% PSNR 0.9996/0.9973 /0.3384 33.07 0.9994/0.9628 /0.2218 24.29 0.9943/0.9867 /0.2485 31.35 0.9995/0.9972 /0.2935 32.92 Jpeg 30% PSNR 0.9987/0.9956 /0.3373 32.34 0.9981/0.9618 /0.2230 23.87 0.9934/0.9827 /0.2487 30.83 0.9988/0.9955 /0.2910 32.16 Jpeg 10% PSNR 0.9895/0.9657 /0.3288 29.61 0.9849/0.9311 /0.2244 22.48 0.9791/0.9561 /0.2419 28.78 0.9890/0.9732 /0.2873 29.33 Gaussian filter PSNR 0.9987/0.9932 /0.3386 32.05 0.9971/0.9597 /0.2209 23.47 0.9919/0.9769 /0.2466 30.20 0.9970/0.9892 /0.2925 30.70 Median filter PSNR 0.8734/0.7536 /0.3367 25.39 0.9792/0.9214 /0.2153 22.83 0.9536/0.9119 /0.2499 28.25 0.9754/0.9400 /0.2828 28.23 Sharpening PSNR 0.9877/0.9461 /0.3365 23.25 0.9624/0.8833 /0.2138 16.44 0.9689/0.9201 /0.2526 22.71 0.9854/0.9330 /0.2868 23.44 FMLR PSNR 0.9448/0.8149 /0.3288 30.67 0.9836/0.8874 /0.2215 24.57 0.9229/0.7814 /0.2115 28.97 0.9655/0.8831 /0.2784 30.87

51 Enhanced watermarking by repetition codes

52 Performance (basic vs. enhanced) By soft-decision:  The difference between basic and enhanced scheme is small. By hard-decision:  Enhanced scheme is superior to basic scheme. (except some particular attacks)

53 OUTLINE Introduction to watermarking SPIHT Factors that affect watermarking system Proposed watermarking scheme Conclusion

54 Our watermarking system suffers less from the host interference and quantization noise. The embedding and extracting of watermarks are simple. The performance can be improved by enhanced scheme with larger noise margins.

55


Download ppt "Bit-Plane Watermarking for SPIHT-Coded Images 台北科技大學資工所 指導教授:楊士萱 學生:廖武傑 2003/07/29."

Similar presentations


Ads by Google