SVD and chaotic system based watermarking approach for recovering crime scene image

This paper presents a watermarking approach for recovering crime scene image when the crime scene image has been tampered, based on chaotic system and singular value decomposition (SVD). At first, authentication information has been embedded into the region of information (ROI) of the scrambled crime scene image. The authentication information has been generated by using logistic map, while the scrambled crime scene image has been obtained by two dimensional Arnold´s cat map. After then, ROI has been compressed by discrete cosine transform (DCT) and the compressed ROI has been embedded into region of background (ROB) by using singular value decomposition. In extraction stage, the authentication information has been extracted from the ROI of the watermarked crime scene image. If the extracted authentication information is not same as the embedded information, ROI has been extracted from the ROB. The results of the proposed method has been evaluated by using peak signal to noise ratio (PSNR) and mean squared error (MSE) performance parameters. At the same time, the proposed watermarking approach has also been compared with other methods. The proposed method gives better values of MSE and PSNR than the compared methods.