Face super-resolution reconstruction method capable of recovering identity information

A technology of super-resolution reconstruction and identity information, which is applied in the field of face super-resolution reconstruction that can recover identity information, can solve the problem of not recovering identity features well, achieve high accuracy, improve performance, and simple training steps Effect

Pending Publication Date: 2022-02-18
SOUTH CHINA UNIV OF TECH +1
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Because although the existing methods reconstruct a seemingly good high-resolution face image, they do not recover the details of identity features and other information well.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Face super-resolution reconstruction method capable of recovering identity information
  • Face super-resolution reconstruction method capable of recovering identity information
  • Face super-resolution reconstruction method capable of recovering identity information

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0070] The present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.

[0071] Such as Figure 1 to Figure 4 As shown, the face super-resolution reconstruction method of recoverable identity information provided by this example includes the following steps:

[0072] 1) The data set selected for training is CASIA-WebFace, which contains 453453 face image samples from 10575 different identities. After performing preprocessing such as face detection and alignment on the images in the training data set, a batch of 32 images is randomly selected as the high-resolution target image I HR , where each image I HR The size is 112×112×3 (respectively representing the length, width and number of color channels of the input image), and the pixel value is a floating point number ranging from -1 to 1. for each I HR , using the bicubic interpolation alg...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a face super-resolution reconstruction method capable of recovering identity information, which comprises the following steps: 1) preparing training data including a low-resolution input image, a high-resolution target image and a target face mask; 2) constructing a face super-resolution network, wherein the face super-resolution network is composed of a main branch and an auxiliary branch which cooperate with each other; 3) sending a low-resolution image into the network in the step 2) to obtain a high-resolution image and a face mask output by a secondary branch; 4) calculating total loss, wherein the total loss comprises mask pixel loss, face mask loss and perception feature loss; 5) optimizing and updating network parameters according to the total loss in the step 4); 6) verifying whether the network performance in the step 5) reaches the standard or not, and if yes, stopping training; and if not, repeating the steps 1)-6). According to the method, the problem that the recognition rate of the face image generated by the existing face super-resolution reconstruction method is low is solved.

Description

technical field [0001] The invention relates to the technical field of face image reconstruction, in particular to a face super-resolution reconstruction method capable of recovering identity information. Background technique [0002] Face recognition is an identification technology based on facial biometric information. It is widely used in security systems, public places, traffic entrances and exits, community access control and other places of life and work. Compared with biological characteristics such as fingerprints, irises, and voices, face recognition has natural advantages such as non-contact, directness, and concealment, so it is the first choice for identity recognition applications. However, due to the hardware conditions of the imaging equipment and the constraints of the light environment when shooting, the collected face images often cannot meet the high-resolution requirements, which leads to the accuracy rate of the existing face recognition algorithms under...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06T3/40G06T5/20G06T5/30G06N3/04G06N3/08G06T7/00
CPCG06T3/4053G06T5/30G06T5/20G06N3/084G06T7/0002G06T2207/30201G06N3/045
Inventor 田联房孙峥峥杜启亮
Owner SOUTH CHINA UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap