Single-image super-resolution reconstruction method based on deep residual network

Abstract

The invention discloses a single-image super-resolution reconstruction method based on a deep residual network. The method of the invention mainly comprises a first step of performing block extraction and pixel averaging processing on an image in a sample image database to obtain a corresponding high resolution and low resolution training image sets; a second step of constructing a deep convolutional neutral network with a residual structure for iterative training, and then inputting the training set obtained in the first step to the neural network constructed in the second step for iterative training; and a third step of according to a data model obtained by training, realizing the continuous up-scaling of the input low resolution image through the combination of iterative operation and an interpolation algorithm. By introducing a deep residual network and introducing an upsampling layer at the end of the network, the method of the invention accelerates the processing speed of the image up-scaling, enhances the display effect of the image details, obtains a better image super-resolution reconstruction effect, and has a wide range of applications in the image high definition display, image compression, security checks and other fields.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a single image super-resolution reconstruction method based on a deep residual network. Background technique [0002] At present, with the widespread application of multimedia technology and the widespread popularization of digital equipment, digital images have been widely used in various fields due to their advantages such as low cost, good real-time performance, and convenient post-processing. However, limited by the physical size and number of sensors, the spatial resolution of the digital image generated by the imaging system is often difficult to meet the needs of the reviewers, and limited by the imaging principle and manufacturing process, manufacturing an imaging system with high spatial resolution will be difficult. It will greatly increase the cost and development cycle of the system. Therefore, under the condition of the same hardware system, it ...

AI Technical Summary

Problems solved by technology

[0006] In view of the shortcomings of some existing super-resolution technologies such as low reconstruction accuracy, slow reconstruction speed, and inability to continuously scale up images, etc.

Aiming at the defect of low reconstruction accuracy of the existing method, the present invention introduces the residual structure into the neural network, eliminates the gradient dispersion phenomenon in the neural network, greatly increases the depth of the neural network, and improves the speed of the training process and the accuracy of the reconstructed image

Aiming at the defect of slow reconstruction speed in existing methods, the present invention introduces an upsampling layer at the end of the neural network, and reduces the number of parameters in the neural network, thereby improving the speed of image reconstruction

Aiming at the disadvantage that the existing method cannot continuously enlarge the image, the present invention, in the process of image reconstruction, iteratively processes the low-resolution image through the neural network, and uses the interpolation algorithm to adjust its size so as to realize the continuous scaling of the image

Images