High-Resolution Image Reconstruction Method Based on Low-Rank Tensor and Hierarchical Dictionary Learning

Abstract

A high-resolution image reconstruction method based on low-rank tensors and hierarchical dictionary learning: use bilinear interpolation to upsample and downsample the original image, and use the processing results and the original image as a dictionary learning training set; train the original and downsampled image, extract the downsampled image gradient, arrange the original image and the downsampled gradient into a tensor, and perform low-rank approximation on the latter; perform sparse dictionary learning on the original tensor and the approximate downsampled gradient tensor to obtain the image recovery dictionary; training is low resolution and upsampling images, extract the low-resolution image gradient, arrange the low-resolution gradient and upsampling images into tensors, learn and update the dictionary; transfer the original image to YCbCr space, use the dictionary to reconstruct Y, Cb and Cr Bilinear interpolation is used for reconstruction to obtain the original restored image; iterative back-projection globally enhances the original restored image to obtain the final result. The present invention uses tensors to retain the structural information of the image and improve the accuracy of image reconstruction.

Description

technical field [0001] The invention relates to the field of high-resolution image reconstruction. In particular, it concerns a method for high-resolution image reconstruction based on low-rank tensor and hierarchical dictionary learning. Background technique [0002] The reflection of the human eye on the objective world through various observation systems is called an image. People perceive shape, size, position, distance, etc. through vision, and make corresponding judgments. The development of society has made people's requirements for obtaining high-resolution digital images more and more prominent. Whether it is for military or civilian use, how to obtain a high-resolution digital image has become a problem that people must solve. [0003] The so-called high-resolution digital image refers to a digital image with dense spatial distribution, that is to say, the image has more pixel sets per unit space. For example, medical CT images can be used as the basis for docto...

AI Technical Summary

Problems solved by technology

However, although the image processed in this way can increase the number of pixels in the unit space, in essence, it does not break through the information content of the original image, but only improves the visual effect of the image, and does not increase The amount of information in the image

Images