Image super-resolution method based on clustering regression

Abstract

The invention discloses an image super-resolution method based on clustering regression. The image super-resolution method is specifically implemented according to the following steps: step 1, selecting information features capable of reflecting a pixel similar structure in a low-resolution image; 2, performing clustering segmentation on the information features selected in the step 1 by using a super-pixel segmentation algorithm, and segmenting the image into K classes; 3, respectively learning the dictionary of each class of the K classes obtained in the step 2; 4, determining an optimal base vector; 5, estimating high-resolution pixels through a non-local dictionary regression model on the basis of the optimal base vector selected in the step 4; 6, converting the high-resolution pixel regression estimated in the step 5 into a global optimization unified regular term; and step 7, iteratively optimizing the high-resolution image, and outputting the image. According to the method, details and edges of the image are clearer by learning structural rules of a group of local dictionaries of the low-resolution image, and the problem of super-resolution reconstruction is solved through regularization terms.

Description

technical field [0001] The invention belongs to the technical field of computer vision and relates to an image super-resolution method based on cluster regression. Background technique [0002] Image super-resolution reconstruction is one of the most important research topics in the field of computer vision. At present, due to the limited imaging capability of the equipment and the complex imaging environment and other factors, the quality of the obtained images is usually poor, which cannot meet the requirements of high-resolution applications in practical applications. In the past decades, many different super-resolution image reconstruction methods have been proposed. According to the principle of reconstruction, it can be roughly divided into the following three types: image super-resolution algorithm based on interpolation, image super-resolution algorithm based on reconstruction and image super-resolution algorithm based on instance learning. [0003] The image super...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a method for image super-resolution based on clustering regression, which solves the problem that it is difficult to restore high-frequency information in images in existing reconstruction-based image super-resolution algorithms

Images