Image Restoration Algorithm Research Based On PDE Method

Image processing is an important research subject. The majority of human information is obtained from human vision, and vision information is obtained from images and media, so a lot of clear images are very important for human study and research. The traditional recovery methods process image by filtering out the high frequency components in images, which can achieve good results, but it is likely to lose the image edge and texture features when removal the noise. The research of Partial Differential Equations provides new methods for image processing, and shows great vitality.The quality evaluation of image processing methods depends on the quality of the image which had been processed, so a reliable quality evaluation criteria is essential. Against at unreality of reference picture in reference image quality evaluation criteria, this paper has researched no-reference digital image quality evaluation based on perceptual masking, analyze the image pixels which reflect the characteristics of feature information, combine the concept of anisotropy in partial differential equations, a non-reference image quality evaluation criteria based on anisotropic is proposed. The feasibility and correctness of the new method has been validated by experiments from subjective and objective view point. The new method has great value in the field of no reference images evaluation and can provide a reliable evaluation criteria when there is no reference image in real world.The filtering algorithm based on P-M equation also has been studied in this paper. The coefficients and the nature of this algorithm have been considered, and the reliability and efficiency of the algorithm also have been analyzed. And then the algorithm is implemented in the form of discrete. At last this method is compared with the traditional methods, which proved that it has high efficiency and fastness. Based on P-M equation, the nature and characteristics of diffusion coefficient have been researched. At last the paper summarizes the basic nature of the function of P-M equation. A new diffusion function is presented in this paper which can improve the effectiveness of image filtering method. Experimental results show that the new diffusion function has better performance.