Study, Based On Image Content Authentication Watermark Technology

Along with the fast development of the Internet, a large number of multi-media information transmitted over the network, the authenticity and reliability become serious problem for the multi-media data. In order to judge the accuracy and integrity of the images, authentication techniques have been developed accordingly. In this paper, we describe the general framework, classification and design requirements of an image authentication system. Then, we analysis the advantages and disadvantages about fragile and semi-fragile watermarking. Summarized the main problems of authentication watermarking.Mainly works as following:1. A semi-fragile watermarking algorithm for image authentication is proposed in this paper. Edge-character is extracted by finding the maxima of a wavelet transform modulus, low frequency character is generated by wavelet transform and as another watermark after chaotic. One watermark is used for indicating the modified location and another is used for checking content modification. Watermarks are embedded in the coefficients of middle frequency of wavelet transform. Image authentication and watermark extract needed no information about the original watermark. Experimental results show that the proposed algorithm provided good classification of intentional and incidental tampering.2. Proposed an algorithm that is based on the content of images in the contourlet domain. Modified the HVS model in the wavelet domain to adapt contourlet transform and extracted the watermark. Watermark is embedded in the largest energy sub-band after contourlet transform, increased transparency and security of watermark. Image authentication and watermark extract needed no information about the original watermark and original image, achieved a blind certification. Experimental results show that the proposed algorithm can find the location of tampering. The false alarm point was removed by filtering.3. A zero-watermarking algorithm was proposed based on Nonsubsampled contourlet transform. Edge-character is extracted by Nonsubsampled contourlet transform, the algorithm reduce the interference of noise on the image edge. After the original edge-character was decoded, compared it with the new edge-character extracted from the image that should be certified. Experimental results show that the proposed algorithm has a stronger denoise performance than the traditional edge detection algorithm and can preserve the details of the image edge. For the image have been modified,the certification results can be used to find the location of tampering.