Face super-resolution method based on prior information and attention fusion mechanism

Abstract

The invention discloses a face super-resolution method based on prior information and an attention fusion mechanism. The method comprises the following steps: constructing a training set and a test set; inputting the training set into a rough super-resolution network for processing to obtain ISR1; respectively inputting the ISR1 into an encoder network and a prior information extraction network to obtain a feature graph f and an analysis graph p; inputting the feature graph f and the analytic graph p into a feature fusion network for fusion to obtain fFusion; the fFusion is input into a decoder network to be decoded, and a final result ISR is obtained; constructing a joint loss function, continuously iterating to minimize the loss function, and training to generate a super-resolution network model; according to the method, the problem of insufficient use of face prior information is solved, the attention mechanism is utilized to fuse the feature map and the analytic map, the analytic maps and the feature maps corresponding to different face components are fused respectively, the guiding effect of the analytic maps on the super-resolution of the face image is increased, the reconstruction efficiency is improved, and the reconstruction effect is enhanced.

Description

technical field [0001] The invention belongs to the technical field of digital image processing methods, and relates to a face super-resolution method based on prior information and attention fusion mechanism. Background technique [0002] Image super-resolution is a very important research problem in the field of computer vision and image processing, and the application of image super-resolution reconstruction technology on face images is called "face hallucination (Hallucination)" or face super-resolution (SR) , refers to the super-resolution problem specific to the domain of face images. In many practical situations, limited by physical imaging systems and some human factors, face images are always of low quality. These images often have low resolution and poor recognizability, which hinders communication, criminal investigation, security enhancement, etc. Therefore, face super-resolution has important research significance. With the development of deep learning technol...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to provide a face super-resolution method based on prior information and attention fusion mechanism, which solves the problem of insufficient use of face prior information existing in the prior art, and effectively improves the face image at the same time. Quality of super-resolution reconstructions, including PSNR and SSIM

Images