Initial estimation improvement-based image super-resolution reconstruction method

Abstract

The invention belongs to the fields of image processing and computer vision detection, and relates to an initial estimation improvement-based image super-resolution reconstruction method. The method comprises the steps of shooting similar but not completely same image sequences of a same scene by utilizing a camera, calculating transformation matrixes, relative to a reference image, of other images by taking a first image as the reference image, reserving sub-pixel displacement between the images, performing reverse transformation on the image sequences, and interpolating and amplifying the reference image by 4 times to serve as initial estimation of a high-resolution image; and correcting the initial estimation by utilizing the image sequences subjected to the reverse transformation to obtain a corrected high-resolution image, performing iterative correction on each pixel point of the image by utilizing a convex set projection algorithm until an iterative condition is met, and finally obtaining a reconstructed high-resolution image. According to the method, the pixels of the high-resolution image are properly improved before being subjected to projection correction, so that mean square errors of reconstruction are reduced and higher image reconstruction quality is achieved.

Description

technical field [0001] The invention belongs to the field of image processing and computer vision detection, and relates to an image super-resolution reconstruction method with improved initial estimation, specifically refers to the method of improving the image reconstruction quality by improving the initial estimation of super-resolution images based on a convex set projection algorithm method. Background technique [0002] With the development of science and technology, images have become an important carrier of information transmission, and the demand for high-resolution images is becoming more and more extensive. In remote sensing mapping, medical diagnosis, military reconnaissance and other fields, high-resolution images are required to obtain more detailed information, so as to make more accurate judgments. However, because the camera CCD array density has reached a technical bottleneck, it is difficult to further improve the image resolution, so obtaining higher res...

AI Technical Summary

Problems solved by technology

The above two methods can effectively improve the resolution of the image, but because the convex set projection method is heavily dependent on the accuracy of the initial estimate, simply adopting the interpolation method to obtain it will affect the quality of the final image

Images