A Novel Fast Self-restoration Semi-fragile Watermarking Algorithm for Image Content Authentication Resistant to JPEG Compression Hui Wang, Anthony TS Ho and Xi Zhao 24 th October 2011
Outline Background Image Content Authentication and Self-Restoration Problem? How to solve? Proposed Watermarking Method Watermark Generation, Embedding and Extracting; Image Authentication; Image Restoration Experimental Results and Analysis Conclusion and Future Work 02/18
Background The Problem: Solution Approach: Fragile Semi-fragile Watermarking techniques for Authentication Self-restoration Technical Requirements: 1.Imperceptibility 2.Robustness 3.Security 03/18
Background Generated Watermark Extracted Watermark Watermarked Image Authentication & Restoration 04/18
Background The Problem: Solution Approach: Fragile Semi-fragile Watermarking techniques for Authentication Self-restoration Technical Requirements: 1.Imperceptibility 2.Robustness 3.Security 03/18
Proposed Watermarking Method 1.Embedding Method: Mapping Function : From location(i,j) to location (i, j ) 05/18
Method for embedding and detecting authentication watermark: Quantization Index Modulation (QIM) [16] [16] B. Chen and G. W. Wornell, "Quantization index modulation: a class of provably good methods for digital watermarking and information embedding", IEEE Trans. Inf. Theory, vol. 47, no. 4, pp.1423-1443, 2001. 06/18
Proposed Watermarking Method 2.Authentication and Restoration Method: 07/18
Proposed Watermarking Method Generate watermark for Restoration Use the extract watermark for Restoration 08/18
Proposed Watermarking Method Linear Regression Analysis: 09/18
Proposed Watermarking Method 10/18
Proposed Watermarking Method Probability Distribution Mean Pixel Value AC Coefficient Li et al. s method [12] for image restoration [12]. Li, G., Pei, S., Chen, G., Cao, W.,Wu, B.: A Self-embedded Watermarking Scheme Based on Relationship Function of Corresponding Inter-blocks DCT Coefficient. In: CSCWD 2009, pp. 107-112 (2009) 11/18
Experimental Results Original Image Watermarked Image Modified&Attacked Image Authenticated Image Restored without Attack Restored with JPEG 75 Attack 12/18
Experimental Results Test image: greyscale images size:512x512 pixels Payload: 6bits/ 8*8 block for authentication 4 decimals/ 8*8 block for recovery Imperceptibility PSNR: 37.61 db False Rate of Authentication QF=85 QF=75 Robustness of Restoration Scheme QF=65 13/18
Experimental Results Imperceptibility Quality of Recovered Image [12]. Li, G., Pei, S., Chen, G., Cao, W.,Wu, B.: A Self-embedded Watermarking Scheme Based on Relationship Function of Corresponding Inter-blocks DCT Coefficient. In: CSCWD 2009, pp. 107-112 (2009) [13]. Chamlawi, R., Khan, A., Idris, A.: Wavelet based image authentication and recovery. Journal of Computer Science and Technology, vol. 22(6), pp. 795-804 (2007) 14/18
Experimental Results Robustness of Restoration Scheme Our Methods: Li s [4]: Chamlawi s [5]: JPEG Compression QF=95 QF=85 QF=75 QF=65 15/18
Conclusion and Future Work Conclusion A novel semi-fragile watermarking scheme for: -tampered area localisation -self-restoration Approaches utilised to analyse the feasibility of proposed method: - Linear Regression - Probability distribution Result of data analysis: -The imperceptibility of our watermarked image: 37.61 db -PSNR of restored image on average: 24.71 db -PSNR of restored image after JPEG compression: 2.5 db higher than the other two methods 16/18
Conclusion and Future Work Future Work - test robustness to some other non-malicious attack, such as noise and spatial filtering - investigation new method by merging of watermarks for authentication and self-restoration together 17/18
Hui WANG 26BB02 Computing Department University of Surrey Guildford, UK GU2 7XH 0044-07411098590 h.wang@surrey.ac.uk 18/18