Image super-resolution reconstruction method based on deep convolution sparse coding

Abstract

The invention belongs to the technical field of image super-resolution reconstruction, and discloses an image super-resolution reconstruction method based on deep convolution sparse coding, and the method comprises: embedding a multilayer learning iteration soft threshold algorithm ML-LISTA related to a multilayer convolution sparse coding model ML-CSC into a deep convolution neural network DCNN; adaptively updating all parameters in the ML-LISTA by using the learning ability of the DCNN, and constructing an interpretable end-to-end supervision neural network SRMCSC for image super-resolution reconstruction; and introducing residual learning, extracting residual features by using an ML-LISTA algorithm, combining the residual and an input image to reconstruct a high-resolution image, and then accelerating the training speed and the convergence speed. The SRMCSC network provided by the invention is compact in structure, has good interpretability, can provide a result with visual attraction, and provides a practical solution for super-resolution reconstruction.

Description

technical field [0001] The invention belongs to the technical field of image super-resolution reconstruction, and in particular relates to an image super-resolution reconstruction method based on deep convolution sparse coding. Background technique [0002] Currently, image super-resolution reconstruction (SR) is a classic problem in many digital imaging and computer low-level vision, which aims to construct high-resolution images (HR) from single-input low-resolution images (LR), and is widely used It is used in a variety of fields, from security and surveillance imaging to medical imaging and satellite imaging where more image detail is required. This is due to the imperfection of the imaging system, transmission medium and recording equipment, which affects the visual effect of the image. Therefore, in order to obtain high-quality digital images, it is necessary to perform super-resolution reconstruction on the images. [0003] In recent years, image super-resolution re...

AI Technical Summary

Problems solved by technology

But most methods usually have two main problems and disadvantages: First, these methods are generally computationally complex in optimization, making the reconstruction process time-consuming; certain margin

[0006] (1) The existing super-resolution convolutional neural network structure is not interpretable and can only be designed by trial-and-error techniques, which brings difficulties to the improvement of the network structure

[0007] (2) The existing super-resolution convolutional neural network SRCNN relies on the context of small image regions, which is not enough to restore image details

[0008] The difficulty in solving the above problems and defects is: the existing super-resolution reconstruction convolutional neural network structure is not interpretable, and the design of the network structure relies on trial and error, making it difficult to improve; and it depends on the context of small image regions, which is not enough to restore the image detail

Images