NormLV feature based low-rank sparse neighborhood-embedding super-resolution method

Abstract

The invention discloses a NormLV feature based low-rank sparse neighborhood-embedded super-resolution method. The method includes: firstly, structuring NormLV features to subject a neighborhood embedding algorithm to feature enhancement, and selecting neighbors through the features; secondly, grouping training sets to obtain an index set; thirdly, utilizing low-rank sparse neighborhood-embedding algorithm to calculate a weight matrix; fourthly, subjecting the weight matrix to normalization; fifthly, performing linear combination to obtain high-resolution image blocks; sixthly, fusing the high-resolution image blocks to obtain initial high-resolution images; finally combining priori and global constraints of consistency, and utilizing TV and IBP algorithms to further improve quality of high-resolution image reconstruction. Sparse representation and the neighborhood embedding algorithm are combined, and the technical problem that super-resolution reconstruction quality is affected by inconsistency of low-resolution image and high-resolution image neighborhood relations is solved. Clearer and richer texture details and image edges can be restored by the acquired high-resolution images, and the method has better visual effect as compared with other methods.

Description

technical field [0001] The invention belongs to the technical field of image processing, and mainly relates to an image super-resolution method, specifically a low-rank sparse neighborhood embedding super-resolution method based on NormLV features. The high-quality images thus obtained provide a basis for subsequent image processing, analysis and understanding. Great help, can be used in different fields, such as biomedicine, video and multimedia systems, military reconnaissance and other fields. Background technique [0002] With the increasing popularity of Internet applications and the rapid development of mobile communication technology, image super-resolution reconstruction has been widely used in the field of image processing, which can overcome the lack of resolution of imaging systems. It has achieved good results in different fields, such as biomedicine, video and multimedia systems, military reconnaissance and other fields. After nearly 30 years of development and...

AI Technical Summary

Problems solved by technology

However, the traditional super-resolution method based on the neighborhood embedding algorithm has the following problems: 1) The first-order gradient and second-order gradient feature extraction methods can only represent features in the horizontal and vertical directions, and cannot better represent image blocks, which will cause The filtered neighborhood is inaccurate; 2) Due to the inaccuracy of feature extraction, the mapping between low-resolution image blocks and high-resolution image blocks is not a one-to-one linear mapping; 3) Euclidean distance is used to strictly find fixed Linear embedding of k-nearest neighbors can easily lead to under-fitting or over-fitting, and noise samples or external interference samples cannot be excluded when finding the nearest neighbors; , when the size of the training set is large, the algorithm is time-consuming

In summary, the traditional neighborhood embedding super-resolution method is inaccurate in the neighborhood selected by feature representation, searching for k-nearest neighbors cannot exclude noise samples or external interference samples, and calculating reconstruction weights can easily lead to underfitting or overfitting and The algorithm is time-consuming and has some shortcomings

Images