Research Of Key Technology For Digital Image Disguise

Compared to other multimedia like text, digital images including moreinformation play a very important role in information propagation and knowledgeaccumulation. A large number of images have raised new challenges to mediasecurity. To ensure the security of image storage and transmission, a newtechnique named image disguise emerged. Compared with the traditional imagesecurity protection methods, image disguise technique is more covert and secure.However, the technique is short of more profound theoretical research and theexisted algorithms cannot satisfy the various demands of security. Therefore, it isvery necessary to roundly research key techniques related to image disguise andits security, both of which would improve the image disguise technique to ahigher level.In order to promote the performance of image disguise system, we statethree key techniques in image disguise: image steganography, image inpaintingand image synthesis. Meanwhile, we give analysis of the key problems existed inimage disguise and introduce some novel approaches to solve them. Thus, thesecurity and applicability of image disguise system have remarkable advance. Inthis thesis, the main research and contributions are listed as follows:(1) We proposed a new steganography method based on genetic algorithmwhich increases the security of steganography method and preserves a highquality of stego image. From the applicability perspective, we applied affinetransformation to remove the statistical abnormality of the stego image, thus, thedetection resistant ability of steganography is improved.(2) We researched the characteristics of image copy-move detection anddesigned a more secure image inpainting method. The traditional imageinpainting methods only focus on visual effect and haven’t the ability of blindforensics-resistant. The algorithm in this thesis made the copy-moved image isnot only secure but also preserves a good visual quality.(3) Directing at the feature of block compression in JPEG image, weconstructed the filter to remove the inconsistent block artifact grid and ensure theblock consistency in JPEG image synthesis. Meanwhile, directing at image processing such as re-sample and re-compression, we adjusted the statisticalproperty of image via optimization modeling which achieve the statisticalconsistency of images from different sources. The experiments proved theeffectiveness of the algorithms to remove the traces of image synthesis.(4) We designed two systems with high concealment: image transmissionsystem and image Secret Classification Mark system. Furthermore, the propertyof image disguise applied in the two system environments is researched.Experimental results have demonstrated the practicability of the design.