Image Restoration and Matching Integrated Method and System Based on Hierarchical Sparse Representation

Abstract

The invention discloses an integrated method and system for image restoration and matching based on hierarchical sparse representation, belonging to the field of computer vision, including: using the sliding window method to construct an original dictionary; The one-dimensional vector of the first-level dictionary constitutes a first-level dictionary; the one-dimensional vector corresponding to the first-level dictionary and the query image is used for sparse representation to obtain the sparse coefficient α 1 and the reconstructed image x 1 ; will reconstruct the image x 1 As a blurred image, and according to the sparse coefficient α 1 Construct a secondary dictionary with non-zero components in the middle; use the secondary dictionary and fuzzy image for sparse representation to obtain the sparse coefficient α 2 and the reconstructed image x 2 , and update the updated blurred image to the reconstructed image x 2 , repeat this step until the maximum number of iterations is reached; obtain the sparse coefficient α 2 The image block corresponding to the largest component in is used as the matching image to obtain the coordinates of the query image in the reference image. The invention can improve the speed of image matching while ensuring high image matching precision.

Description

technical field [0001] The invention belongs to the field of computer vision, and more specifically relates to a method and system for integrating image restoration and matching based on hierarchical sparse representation. Background technique [0002] Image matching is one of the classic problems in computer vision, which can be widely used in vision-based navigation systems, remote sensing images and other scenarios. The so-called image matching refers to the process of identifying points with the same name (points at the same physical location between different images) from two or more images through a certain algorithm. For example, given a reference image and a query image, the coordinates of the query image in the reference image can be determined through image matching. [0003] Most of the image matching algorithms assume that the input query image is in an ideal state by default, but in practical applications, the obtained query image is usually an image with real-...

AI Technical Summary

Problems solved by technology

[0004] JRM-DSR can effectively solve the problem of low matching accuracy caused by the degeneration of the query image. However, when the size of the reference image is much larger than the size of the query image, the size of the dictionary extracted from the reference image will decrease. becomes very large, which leads to the need for a lot of calculations when using a dictionary matrix for sparse representation, so that the speed of image matching is very slow

Images