Single image super-resolution reconstruction method guided by self-paradigm learning

Abstract

The invention provides a single image super-resolution reconstruction method guided by self-paradigm learning. By introducing the geometric transformation of the image blocks, the retrieval space of internal similar image blocks is expanded, so that the target image blocks can retrieve more image blocks with higher similarity in the image pyramid, the high-frequency detail information in the reconstructed image obtained by the algorithm is more sufficient, and the reconstruction effect on the internal texture of the image is better; the internal plane information of the image is acquired by adopting an evanescent point detection method, and the retrieval area is constrained by utilizing the local relevance of the image in the process of retrieving similar image blocks, so that the calculation amount of nearest neighbor search retrieval is greatly reduced, and the algorithm efficiency is improved. Compared with the method in the prior art, the image quality reconstructed by the method is greatly improved, the recovery quality of the internal texture and edge of the image is high, and the objective evaluation index of the reconstructed image is also obviously improved.

Description

technical field [0001] The invention relates to a method for super-resolution reconstruction of a single image, in particular to a method for super-resolution reconstruction of a single image guided by learning from examples, and belongs to the technical field of image super-resolution reconstruction. Background technique [0002] The resolution of an image is a measure of the amount of information contained in an image, describing the number of pixels in a unit space. In order to process the image more effectively and make the image more widely and efficiently used, it is generally hoped that the quality of the image to be processed should be as high as possible in actual use. In terms of observation and user's subjective experience, high-quality images have more detailed information than low-quality images and it is easier to distinguish the edges of images. By increasing the resolution of images, the number of pixels per unit area increases, and the image interior The de...

AI Technical Summary

Problems solved by technology

However, due to the large number of complex texture features in the original image, the self-example cannot reconstruct all the details of the image well due to the limited number.

The loss of image detail information is serious, the search space of internal image blocks is small, the matching of multi-scale similar image blocks is too large, and the similarity between matched image blocks is low

[0014] Fourth, the PatchMatch algorithm in the prior art uses random or prior information to initialize the nearest neighbor search space. For the solution space of the 7-dimensional geometric transformation of the image block, the convergence efficiency of the random initialization method is low, and local optimal solutions cannot be avoided. situation

Images