Research On Image Hashing Algorithm Based On Chaotic System

With the fast development of online communication and network, multimedia especially digital image has become an important way of getting information, and also it has been an important part of our daily life. However, with the development of the technology of the processing of image, it is easy for people to change and edit the image and that is invisible to eyes. So, a big problem has come out that is the facticity and integrity of the images. On the other hand, the requirement of search for images is becoming larger and larger with the the rapid expansion of images. As one of information security technology,image hashing can turn different size pictures into a robust hash sequence.It is used in the fields of image retrieval,image authentication and copyright protection,etc.The research on image hashing algorithm attracts the interest of scholars all over the world.At the same time, the Chaotic System has been widely applied to the image encryption because of its excellent cryptographic properties,this paper combines the image hashing with the Chaotic System to enhance its security.After studying current image hashing algorithms in depth, taking gray image as fundamental research object, two image hashing algorithms are proposed. Firstly, Difference Quantization of Histogram (DQH) based on image grayscale compression is studied; together with Chaotic System, the image hashing algorithm based on Difference Quantization of Histogram and Chaotic System is proposed. To begin with, the image is compressed in grayscales. After that, intermediate image hash is obtained form modulation between chaotic sequence and the probability sequence of every grayscale of the compressed image. At last, final image hash is generated by binarization and DQH. It is very easy and fast. Secondly, after researching on overlapped image blocks, Discrete Wavelet Transformation (DWT) and Singular Value Decomposition (SVD), a more robust image hashing algorithm is proposed, it's also based on the chaotic system. Firstly, the overlap-intersected image blocks are transformed with wavelet, and low frequency wavelet coefficients are extracted. After that, the standard deviation is obtained from one wavelet coefficient with SVD; it is 1-bit intermediate image hash. At last, final image hash is generated by binarization and DQH.Experimental results show that the proposed schemes are robust against JPEG compression, low-pass filtering, gauss-noise, scaling and rotation attacks, moreover, the second one is better than the first one. Both of them are sensitive to malicious tamper, and probability of conflict among different image hashs is low. Additionally, the Chaotic System obviously enhances the security, DQH enhances the robustness and the binarization by the median of the sequence ensures that 0 and 1 has equal probability in the final image hash.