The Research Of Steganography Based On Domain Of DCT

In recent years, computer network technology and multimedia technology has made rapid progress and wide application,at the same time, network information security and protection problem has become an important issue to be solved. Writing for the number of hidden as an important means of covert communication, has become a new hotspot in the field of information security in recent years. Steganography hides the communication process mainly through the "presence " to protect the security of communications.This paper analyzes the existing model of steganography and classification, and evaluation of the figures from the horizontal perspective of steganography algorithms in the study of the current part of the evaluation method, based on the combination of several key steganography objective evaluation criteria-Hide capacity, peak signal to noise ratio and the DCT coefficient histogram features.This paper studies the JPEG quantization table and the human visual model and the introduction of human visual model based on the definition of quantization table, and then propose a steganography idea-hide the secret message in the quantization table, rather than the DCT coefficients, thus hiding Algorithm is relatively simple and fast to implement. In the extraction of secret information, do not need to be carried out entropy decoding to obtain the quantized DCT coefficients as traditional methods, but the direct extraction of secret information for the quantization table, so this method is simple. Experimental results show that using a custom quantization table hiding secret information, you can achieve better PSNR, can better maintain the DCT histogram features to ensure the security of steganography; but the capacity of quantization table determines it only for hiding a small amount of information.In this paper, from the perspective of improving the capacity of steganography, combine with QTM algorithm to make adjustments to a custom quantization table, increase the non-zero quantized DCT coefficient domain, provides the conditions to enhance the hiding capacity. The main idea of the algorithm is to reduce the frequency coefficient appropriate, and then use the DCT coefficient quantization, the quantized DCT coefficient is higher than the original DCTcoefficient, frequency domain non-0 coefficient increases significantly, on this basis, the proposed rules embedded information embedded in the DCT domain. Experimental results show that the algorithm has a relatively large increase on the hiding capacity while maintaining the peak signal to noise ratio of JPEG images and the DCT histogram features.