Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

A face image illumination recognition method and system based on multi-feature fusion

A multi-feature fusion, face image technology, applied in the field of image recognition research, can solve the problems of low model accuracy, algorithm accuracy can not meet the requirements of illumination discrimination, underfitting and other problems, achieving fast running speed and convenient deployment , the effect of high accuracy

Active Publication Date: 2019-04-05
SUN YAT SEN UNIV
View PDF11 Cites 9 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0011] The naive Bayesian method is the simplest machine learning algorithm, which is fast in classification and easy to understand, but the accuracy of the model is generally low, and it is prone to underfitting
The algorithm accuracy of the decision tree method cannot meet the requirements for illumination discrimination
Existing classification methods have problems such as inaccurate classification, long time-consuming, and complex calculations.
[0012] The method of deep learning has higher accuracy, but the model is more complex and depends on the robustness of the previous training model

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
  • A face image illumination recognition method and system based on multi-feature fusion
  • A face image illumination recognition method and system based on multi-feature fusion
  • A face image illumination recognition method and system based on multi-feature fusion

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0051] Such as figure 1 As shown, this embodiment is a face image illumination recognition method based on multi-feature fusion, including a training phase and an actual recognition phase, which will be described in detail below.

[0052] 1. Training stage

[0053] A training set with 16 lighting conditions was selected from the Multi-PIE database to train the auto-encoder model and the FaceNet model. Model training can be implemented using existing technologies.

[0054] Second, the actual identification stage

[0055] combined with figure 1 , 2 , mainly including the following steps:

[0056] S1. Calculate statistical features for the face image under actual lighting conditions.

[0057] First, use opencv's calcCovarMatrix to extract the covariance matrix of the image, straighten the matrix, and use the trained auto-encoder model to compress the features to 125 dimensions to obtain 125-dimensional image features.

[0058] Then, find the average luminance difference be...

Embodiment 2

[0068] Such as image 3 As shown, this embodiment discloses a multi-feature fusion facial image illumination recognition system, including:

[0069] The statistical feature extraction module is used to extract the covariance matrix of the image for each face image under different lighting conditions, calculate the regional difference feature of the image, and fuse the two as the statistical feature of the image;

[0070] The deep feature extraction module is used to extract the deep feature of the face image based on the neural network method;

[0071] A feature fusion module is used to fuse the statistical features and depth features to obtain the fused features;

[0072] The classification module is used to classify the fused features to realize the illumination recognition of the face image.

[0073] In order to process the actual image, the system also establishes a pre-training module for training the model to generate the auto-encoder model and the FaceNet model that t...

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 multi-feature fusion face image illumination recognition method, which comprises the steps of extracting a covariance matrix of an image for each face image under different illumination conditions, calculating regional difference features of the image, and fusing the covariance matrix and the regional difference features as statistical features of the image; Extracting depth features of the face image based on a neural network method; Fusing the statistical features and the depth features to obtain fused features; And classifying the fused features to realize illumination recognition of the face image. The invention further discloses a face image illumination recognition system based on multi-feature fusion. According to the method, the face images under differentillumination conditions can be distinguished, the accuracy is high, the recognition speed is high, and the future technology is easy to update.

Description

technical field [0001] The invention relates to the field of image recognition research, in particular to a multi-feature fusion face image illumination recognition method and system. Background technique [0002] With the development of electronic technology, the acquisition of images has become more and more convenient. As a carrier of information, images are more visual and intuitive than text and sound. In the field of intelligent security, face recognition technology has attracted widespread attention and has broad application prospects. However, in the process of collecting face images, there are factors such as face illumination, which will reduce the accuracy of face recognition. Therefore, how to accurately distinguish face images under different lighting conditions from a large number of face images, so as to select images suitable for the lighting conditions of the scene under different demand scenarios is of great research significance. [0003] The traditional ...

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): G06K9/00G06K9/62
CPCG06V40/172G06V40/168G06F18/2411
Inventor 赖剑煌吴卓亮欧阳柳谢晓华
Owner SUN YAT SEN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com