Super-resolution image reconstruction method for robust

Abstract

The invention discloses a super-resolution image reconstruction method for robust, and belongs to the technical field of image processing. The method comprises the following steps of: (1) amplifying the interpolation of each input original image; (2) performing motion estimation on the image to be reconstructed and other images in the images amplified in the step (1) to obtain motion estimation parameters; and (3) establishing a super-resolution image reconstruction model considering the motion estimation errors according to the motion estimation parameters obtained in the step (2), and obtaining the reconstructed high-resolution image by solving. Compared with the prior art, the method has the advantages that: 1) an error correction function is added in a framework of the traditional super-resolution image reconstruction method, so that the super-resolution image reconstruction result of the method is superior to that of the traditional method; and 2) proper parameters are adaptivelyselected according to the magnitude of the errors, and the proper parameters can be automatically selected when the errors are great or small, so the method is superior to a method using a fixed parameter all the time.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a robust super-resolution image reconstruction method. Background technique [0002] Image resolution has always been an important indicator for people to measure image quality. Since 1970, with the invention of CCD (Charge Coupled Device) and CMOS (Complementary Metal Oxide Semiconductor) image sensors, the quality of people's acquired images has reached a new level. However, when the image sensor collects images, it is easily affected by factors such as blur, jitter, undersampling, noise, etc., so it is difficult to further improve the image quality. People have been exploring ways to increase the resolution of images for a long time. Traditionally, people improve the image resolution by improving the image sensor, but this method is too expensive and difficult to popularize. Therefore, it has been proposed to use super-resolution image reconstruction technology t...

AI Technical Summary

Problems solved by technology

[0004] However, the objective function of traditional super-resolution image reconstruction has the following defects: first, the error term uses the L2 norm to expand the impact of motion estimation error; second, the error size of the image sequence after motion estimation is not the same, Different weights should be assigned, but the objective function of traditional super-resolution image reconstruction assigns the same weight to each image sequence

Images