Research On Steganography Of Multiple Cover Images

The digital image steganography technology embeds the secret message into digital image through specific algorithms to realize covert communication.The existing digital image steganography technology is mainly based on a single cover image for steganography.However,with the rapid development and wide application of multimedia technology and social network,the behavior of uploading and downloading images in batch becomes very common.Therefore,the multiple cover images steganography technology has attracted the attention of researchers in recent years.Multiple cover images steganography has the characteristics of multiple images and embedding a large number of messages,so the main problem faced by multiple cover images steganography technology is how to distribute messages among multiple images reasonably so that the secret images are the most difficult to be detected.In this paper,aiming at the problem of payload distribution in multiple cover images steganography,three multiple cover images steganography algorithms are proposed from the perspective of improving embedding efficiency and anti-detection.The main works are as follows:The existing typical multiple images steganography algorithm finds the safe capacity of the cover images by iteratively replacing the embedding rate,and the payload distribution efficiency is low.For this reason,a JPEG image multiple cover images steganography algorithm based on dynamic programming is proposed.First,the secret message is averagely embedded into the multi-carrier image,and the features of the cover image set and the secret image set after the average embedded message are extracted;and then calculate the maximum average difference between the features of the cover image set and the secret image set,and use the features of the cover image combined with the idea of dynamic programming to redistribute the payload to minimize the maximum average difference,thereby improving the payload distribution efficiency while reducing the detection error rate of the stego image;and finally,the data decomposition mechanism based on Vandermonde matrix is used to solve the problem of loss or damage in the transmission of secret messages.The experimental results show that when combining the algorithm with the ns F5 and J-UNIWARD steganography,compared with the existing feature reverse iterative replacement multiple cover steganography algorithm,the payload distribution time of the proposed algorithm can be increased by 4 times,and at the same time the anti-detection performance of the encrypted image is improved.At present,the multiple images adaptive steganography algorithm selects images with complex texture to embed the secret messages,but the measurement of the texture complexity of the cover images are inaccurate.To this end,a JPEG image multiple images adaptive steganography algorithm based on 2D Gabor filtering is proposed.First,this paper uses 2D Gabor filtering to construct residual images in different directions and scales,and fully extracts the cover image features.Then,the absolute values of different residual images are averaged,and the image distortion of the average residual image is used to measure the complexity of the cover image.Finally,according to the above measurement results of the complexity of the cover images,the images with high texture content are selected as the selected multiple cover images to embed the secret message.The experimental results show that when the proposed algorithm is applied to J-UNIWARD and UED adaptive steganography algorithms,compared with the existing multiple images adaptive steganography algorithms based on image residuals,image energy,and image fluctuations,the anti-detection performance of the cover images selected by the proposed algorithm and combined with the J-UNIWARD algorithm to generate the stego images is increased by up to 0.8%.The existing multiple images adaptive steganography algorithms based on image residuals and image fluctuations do not fully consider the characteristics of the correlation between spatial image pixels,and the detection resistance of generated stego images needs to be improved.Therefore,a multiple images adaptive steganography algorithm for spatial images based on LBP(local binary pattern)features is proposed.First calculate the LBP feature of the image,and describe the local texture feature of the image more accurately by directly calculating the difference of the image pixels.Then,the entropy of the image is calculated based on the LBP feature image,and the entropy of the image is used to measure the texture complexity of the cover images.Finally,according to the texture complexity of the cover images,the images with complex textures are screened as cover images for multiple cover images steganography.The experimental results show that the proposed algorithm can be combined with WOW and HILL spatial domain adaptive steganography.When combined with WOW algorithm,the proposed algorithm increases by 1.95% compared to the multiple cover steganography method based on image residuals.Compared with the multiple cover steganography based on image fluctuation,it has increased by 1.82%.At last,it summarizes the work of the full text and points out the problems to be studied in the next step.