Research Of Image Fusion Based On Sparse Representation And Chaotic Encryption

With the development of image processing and computer vision,image fusion technology has been rapidly developed and applied,and has become one of the important means of image processing.It mainly combines the images collected by multiple sensors into the same target or scene,and re-obtains more images with source image information.It can help people quickly understand the meaning of the images.In addition,besides studying the development of image fusion technology,it is also important to ensure the security of image information in vulnerable and insecure transmission networks.Based on the theory of filter and sparse representation,in this thesis,an image fusion algorithm is designed by making full use of the dimensionality reduction,denoising and other features as well as the image fusion characteristics.Based on chaos theory and image encryption theory,an image fusion algorithm combining sparse representation and chaotic encryption is designed by making full use of the high sensitivity,irregularity and randomness of chaotic systems.The main work of this thesis includes:(1)Image fusion based on sparse representation,filter theory,chaotic theory and chaotic encryption algorithm are studied.The thesis has a systematic and comprehensive study of how to design efficient image fusion algorithm based on sparse representation and chaotic encryption.By studying the advantages and disadvantages of existing algorithms,the characteristics of the main image fusion and chaotic encryption algorithms are analyzed,which lays a foundation for future research.(2)An image fusion algorithm based on guided filtering and sparse representation is proposed.The algorithm can solve the problem that the transform domain fusion algorithm cannot express the source image information more efficiently and completely,and the problem that single dictionary learning ignores the local features of the image.Firstly,the image is decomposed by multi-scale guided filtering.Then,considering different structural characteristics of images,we use different image fusion rules.For the cartoon structure and the texture structure image with more information,the respective dictionary is trained to perform sparse representation to obtain more accurate sparse coefficients.For the edge structure image with more details,the guided filter weighting fusion rule is used to accurately extract the target edge information,and reduce the computational complexity.Finally,the fused image can be reconstructed.The experimental results show that the fusion scheme effectively integrates the focal region into the fused image and is better than other methods in subjective and objective evaluation indicators.(3)An image security fusion algorithm combining sparse representation and chaotic encryption is proposed.The scheme can guarantee the security of fusion process and achieve high quality fused images.First,the source image is preprocessed,and then the new one-dimensional chaotic system initial value and system parameters are generated by using both the fused detail image and key,and used in the confusion and diffusion steps.They are related to the source image,so the scheme can resist Known and Chosen plaintext attacks.Then,a dynamic diffusion mechanism is proposed.In the encryption process,different states are obtained to encrypt the next bit,and thereby the security of the fusion algorithm can be improved.Finally the fusion image is obtained by decrypting the ciphertext image.In addition,experiments are carried out to analyze the security of the fusion algorithm.The results show that the proposed image fusion algorithmcan not only ensure image security,but also obtain better quality fused images,which has important practical value in secure communication.