Researches On Image Steganography And Steganalysis Algorithm

Image steganography is related to the security of secret communication.The better the image steganography is,the more secure the secret communication is.On the contrary,image steganalysis is related to communication security on the common channel.The better the performance of the image steganalysis algorithm is,the easier it is to detect the existence of secret communication.If the secret communication is detected,necessary measures like cutting off secret communication can be adopted to protect our personal life and national security.Therefore,the main work of this thesis includes the researches of image steganography and image steganalysis algorithm.The main work is as follows:(1)TPSA algorithm,a new image steganographic algorithm,is proposed in this thesis.TPSA is a low-rate and high-payload steganographic algorithm.Three pixels are processed as a unit,the exclusive-or result of the second least significant bits of the pixels in that unit is regarded as the flag.According to the value of the flag,as well as the Hamming distance between the least significant bits and the secret messages,three bits of secret message are embedded.Secret message can be easily extracted depending on the value of flag.Local texture is also considered to achieve adaptive embedding.The advantage of the proposed algorithm is that at most one pixel will be modified when embedding three bits of message into three pixels.(2)FPSA algorithm,a new image steganographic algorithm similar to TPSA,is proposed in this thesis.FPSA is a low-rate steganographic algorithm with four pixels as the group.Four pixels are processed as a unit,in which the exclusive-or result of the second least significant bits of the first three pixels and the least significant bits of the forth pixel is regarded as the flag.According to the value of the flag,as well as the Hamming distance between the least significant bits of the first three pixels and the secret messages,three bits of secret message are embedded.Secret message can be easily extracted depending on the value of flag.Local texture is also considered to achieve adaptive embedding.(3)A new image steganalysis algorithm called B-MTL-CNN based on convolutional neural network is proposed in this thesis.Based on previous researches,this algorithm combines fully convolutional network and multi-task learning,and proposes the concept of image block steganalysis.And the network contains a backbone network and an auxiliary task branch network.The backbone network,whose first layer is initialized with 30 high-pass filters,is used for image steganalysis.And the auxiliary task branch network,which is combined with transposed convolution to upsample the feature maps,is used for image block steganalysis.