Image Inpainting Method Based on Tight Frame Feature Dictionary

Abstract

The present invention proposes an image repair method based on a tight-frame feature dictionary, uses the discrete cosine transform DCT-II type orthogonal matrix to construct a redundant discrete cosine transform DCT frame system, and successfully applies it to the field of image repair. Effectively restore the texture and structure information of images. Under the decomposition of the DCT small frame basis, the frame coefficients obtained represent the edge feature information of different directions or different orders of the image, and at the same time, using the prior knowledge of the sparse frame coefficients, the weighted l 1 Norm DCT frame coefficient optimization model, an iterative algorithm based on approximation operator is proposed to obtain the solution of the model. Under the assumption of the probability model, the Laplace probability distribution prior model is used to approximate the actual probability distribution of the geometric frame coefficients, and the Gaussian distribution of the model noise is assumed, and the MAP technology is used to establish an adaptive sparse soft threshold operator. Sparse representation of geometric frame coefficients can not only protect edge features but also filter out noise.

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