Image super-resolution rebuilding method based on sparse representation

Abstract

The invention discloses an image super-resolution rebuilding method based on sparse representation. The image super-resolution rebuilding method comprises a sample training step and an image super-resolution rebuilding step. The sample building step comprises the step of calculating gradient information of a low-resolution image and residual error information of a high-resolution image and a low-resolution image and the step of acquiring a low-resolution characteristic set and a high-resolution characteristic set through a sparse representation method. The image super-resolution rebuilding step comprises the step of calculating gradient information of the low-resolution image to be processed, the step of finding out the sparse representation coefficient vector from the low-resolution characteristic set and the step of finding out corresponding residual error information from the high-resolution characteristic set, fusing the residual error information into the low-resolution image and acquiring the high-resolution image. The high-resolution image acquired according to the method is richer in detail, clearer in edge, and better in visual effect. The image super-resolution rebuilding method based on sparse representation can be applied to the process of converting standard-definition videos to high-definition videos.

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to an image super-resolution reconstruction method based on sparse representation. Background technique [0002] Standard Definition (SD) refers to VCD, DVD, TV programs and other videos with a resolution of about 400 lines. High Definition (HD) refers to a video with at least 720 lines of progressive scan or 1080 lines of interlaced scan and a screen aspect ratio of 16:9. Comparing the specifications of standard-definition video and high-definition video, we can see that the difference between the two is mainly in resolution and aspect ratio, and the key to conversion is the improvement of video resolution. [0003] Image super-resolution reconstruction technology is a technical solution to improve the resolution of images. Currently, there are three commonly used image super-resolution methods: interpolation-based, reconstruction-based and learning-based methods. The in...

AI Technical Summary

Problems solved by technology

The interpolation-based method mainly uses the strong correlation of adjacent pixels to estimate the information of the interpolation point. This type of method is fast and simple, but the processing effect on the image edge area and texture area is not good, and the details of the reconstruction result are not rich enough.

Reconstruction-based methods use the prior assumptions of the generative model from high-resolution images to low-resolution images to solve the super-resolution problem. The prior assumption is inaccurate, and the reconstruction-based method will degrade severely with the increase of magnification. Both methods will cause poor image details and edge parts

Images