Image inpainting method based on tight frame feature dictionary

Abstract

The invention provides an image inpainting method based on a tight frame feature dictionary. A discrete cosine transform DCT-II-type orthogonal matrix is used for constructing a DCT frame system with redundancy, the system is successfully applied to the image inpainting field, and texture structure information of the image can be effectively inpainted. The image is decomposed under a DCT small frame base, the obtained frame coefficients represent edge feature information of the image in different directions or different orders, a priori knowledge of the sparsity of the frame coefficients is used at the same time, a DCT frame coefficient optimization model based on a weighted l1 norm is built, and an iterative algorithm based on an approximation operator is put forward to acquire the solution of the model. Under the assumption of a probability model, a Laplasse probability distribution priori model is used for approximating the actual probability distribution for geometric frame coefficients, and under an assumed model noise Gauss distribution condition, an MAP technology is used for building an adaptive sparse soft threshold operator, geometric frame coefficient sparse representation is carried out on the image, and the noise can be filtered while the edge features can be protected.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to an image restoration method based on a tight frame feature dictionary. Background technique [0002] With the improvement of computer processing capabilities, using computers to assist humans in completing tasks is becoming more and more intelligent. In daily life, the degree of informatization is getting higher and higher, and digital information technology has been widely used in various fields of society, especially the continuous improvement of the popularity of various mobile electronic devices and wireless networks, followed by various Complicated data and how to analyze and process these data, such as processing the image data obtained by the camera equipment, and modifying certain scenes of the picture. The problem of image inpainting is to study the image intelligent algorithm, automatically repair part of the missing or damaged area information in ...

AI Technical Summary

Problems solved by technology

[0007] As shown in Figure 1(a), the black area is the data missing range; Figure 1(b) is the non-patent literature 1 (Z.Xu and J.Sun, "Image inpainting by patch propagation using patch sparsity", IEEETransactions on Image Processing, vol .19, no.5, pp.1153--1165, 2010), the texture structure of the image is not well filled out; Figure 1(c) is the famous image processing software Adobe Photoshop CC The restoration results obtained by the Content-Aware Fill technology in 2014 can fill in certain textual and structural information of images, but the results do not quite meet the visual requirements of human beings.

Images