Prediction-based reversible data hiding using the difference of neighboring pixels - PowerPoint PPT Presentation

slide1 n.
Download
Skip this Video
Loading SlideShow in 5 Seconds..
Prediction-based reversible data hiding using the difference of neighboring pixels PowerPoint Presentation
Download Presentation
Prediction-based reversible data hiding using the difference of neighboring pixels

play fullscreen
1 / 29
Prediction-based reversible data hiding using the difference of neighboring pixels
240 Views
Download Presentation
teva
Download Presentation

Prediction-based reversible data hiding using the difference of neighboring pixels

- - - - - - - - - - - - - - - - - - - - - - - - - - - E N D - - - - - - - - - - - - - - - - - - - - - - - - - - -
Presentation Transcript

  1. Prediction-based reversible data hiding using the difference of neighboring pixels

  2. Outline • Introduction • Related works • PE expansion method • Proposed scheme • Data embedding • Data extraction and recovery • Experimental results • Conclusions

  3. Introduction (1/3) • Data hiding is referred to as a process to hide some information data into cover media. • The data hiding schemes can be generally classified into two categories: - Irreversible data hiding - Reversible data hiding • Reversible data hiding schemes can be classified into two categories: - Difference expansion scheme (DE-based) - Histogram-based

  4. Introduction (2/3) • Data Hiding sd Cover image Secret data Stego image

  5. Introduction (3/3) • Reversible Data Hiding sd Extract Stego image Cover image

  6. Related work – PE expansion method (1/4) • Thodi et al. proposed an expansion scheme based on prediction error (PE) to embed secret data. • Step 1 : The cover image is divided into several overlapping 2×2 blocks. embedding pixel The pixels of one 2 x 2 block Source: Thodi DM, Rodriguez JJ. Expansion embedding techniques for reversible watermarking. IEEE Transactions on Image Processing 2007;16(3):721–30.

  7. Related work – PE expansion method (2/4) • Step 2 : The prediction value can be calculated by Eqs. (1) and (2): • Step 3 : Compute the prediction error e. • Step 4 : Compute the stego pixel p’.

  8. Related work – PE expansion method (3/4) • Example of embedding procedure. S = 1 160 An example of cover image block.

  9. Related work – PE expansion method (4/4) • Example of extraction procedure. S = 1 160 → (101)2 An example of cover image block.

  10. Proposed scheme – Data embedding (1/13) • Step 1: The cover image is divided into two regions. 2 The LSB of the pixels in the non-embeddable region are used to record the information of pixels with overflow or underflow problems when embedding the secret data. 2 Embeddable and non-embeddable regions in the cover image.

  11. Proposed scheme – Data embedding (2/13) • Step 2: The embeddable region is divided into (W − 3) × (H − 3) overlapping blocks sized 2 × 2and initial block index(BI) is 1. 2 embedding pixel 2 The pixels of one embeddable block.

  12. Proposed scheme – Data embedding (3/13) • Step 3: The absolute difference of the neighboring pixels is calculated by Eq. (3) to identify whether the block is embeddable or not. • Step 4 : The prediction value is calculated by Eqs. (4) : If d < threshold T, Continue. • Step 5 : Compute the prediction error e.

  13. Proposed scheme – Data embedding (4/13) • Step 6: Compute the intermediate stego pixel . • Step 7 : Compute the stego pixel p’. If the pixel cannot be embedded, the BI is transformed into a binary representation and embedded into the non-embeddable region by LSB. • Step 8 : Block index is increased by 1 and repeat the step 3 through 8 until all of the data are embedded.

  14. Proposed scheme – Data embedding (5/13) • Example of embedding procedure. w = {0, 1} , T=2 100 Cover image Stego image 100

  15. Proposed scheme – Data embedding (6/13) • Example of embedding procedure. w = {0, 1} , T=2 Because the absolute difference is higher than the threshold, this block cannot be embedded with the secret data. Cover image Stego image

  16. Proposed scheme – Data embedding (7/13) • Example of embedding procedure. w = {0, 1} , T=2 255 Cover image overflow 255 The stego pixel must be modified into the original pixel, and its block index should be embedded into the non-embeddable region. Stego image

  17. Proposed scheme – Data embedding (8/13) • Example of embedding procedure. w = {0, 1} , T=2 254 Cover image 255 Stego image

  18. Proposed scheme – Data extraction and recovery(9/13) 2 • Example of extraction and recovery procedure. The indices of the overflow and underflow blocks (IOUB) can be extracted from the LSBs of the pixels in the non-embeddable region. T = 2, IOUB = 3 2

  19. Proposed scheme – Data extraction and recovery(10/13) • Example of extraction and recovery procedure. T = 2, IOUB = 3 If BI belongs to IOUB, there is no secret bit in this block. If BI does not belong to IOUB and d is not smaller than T. If BI does not belong to IOUB and d is smaller than T. 254 There is one secret bit in this pixel. → (1)2 Stego image Cover image

  20. Proposed scheme – Data extraction and recovery(11/13) • Example of extraction and recovery procedure. T = 2, IOUB = 3 If BI belongs to IOUB, there is no secret bit in this block. If BI does not belong to IOUB and d is not smaller than T. If BI does not belong to IOUB and d is smaller than T. Stego image The original pixel is the same as the stego pixel. Cover image

  21. Proposed scheme – Data extraction and recovery(12/13) • Example of extraction and recovery procedure. T = 2, IOUB = 3 If BI belongs to IOUB, there is no secret bit in this block. If BI does not belong to IOUB and d is not smaller than T. If BI does not belong to IOUB and d is smaller than T. Stego image There is no secret bit in this block. Cover image

  22. Proposed scheme – Data extraction and recovery(13/13) • Example of extraction and recovery procedure. T = 2, IOUB = 3 If BI belongs to IOUB, there is no secret bit in this block. If BI does not belong to IOUB and d is not smaller than T. If BI does not belong to IOUB and d is smaller than T. 100 There is one secret bit in this pixel. → (0)2 Stego image Cover image

  23. Experimental results (1/5) • The 9 grayscale images of size 512 × 512 (a) Lena (b) Barbara (c) Boats (d) F-16 (e) Pepper (f) Tiffany (g) Sailboat (h)Gold (i)Zeldal

  24. Experimental results (2/5) • The proposed scheme has better performance for smooth images than for complex images. (b) Barbara • Capacity: 214,099 bits • PSNR: 29.4361 dB • T = 43 Cover image Stego image (d) F-16 • Capacity: 257,031 bits • PSNR: 32.7826 dB • T = 255 Cover image Stego image

  25. Experimental results (3/5) • Different thresholds with PSNR values Fig. 1. PSNR values of the proposed scheme with fixed and variant thresholds for the test images.

  26. Experimental results (4/5) • Different thresholds with secret data Fig. 2. Secret data capacity of the proposed scheme with fixed and variant thresholds for the test images.

  27. Experimental results (5/5) Fig. 3. Performance comparison between the proposed scheme and other schemes.

  28. Conclusions • This paper propose a reversible data hiding scheme based on the PE expansion technique. • To use a pre-determined threshold and the absolute difference of the neighboring pixels to determine whether the current pixel is embeddable. • Good image quality. • High hiding capacity.