A deep learning face recognition method based on multi-feature fusion

Abstract

The invention discloses a deep learning face recognition method based on multi-feature fusion. First, two-dimensional gabor transformation (2D gabor) is performed on the orl face database to be tested, and the face database with gabor features is extracted; due to the size of the picture It is: 92×112, the size here is relatively large, and the bilinear interpolation method is used to reduce the image to 32×32; then the original orl face database is fused with the face database of gabor features; finally, the depth is used The stacked self-encoding method in the study is used to encode, and the weight parameters are calculated by softmax regression to predict the recognition accuracy. After the present invention integrates multiple features, the premise is that the accuracy rate of this feature alone can not be lower than 80%, the accuracy rate will be improved, and the algorithm will be more stable, that is, after random initialization, the recognition accuracy rate will basically remain unchanged.

Description

technical field [0001] The invention belongs to the technical field of image recognition and deep learning, and relates to a learning face recognition method, in particular to a deep learning face recognition method based on multi-feature fusion. Background technique [0002] Face recognition is a biometric technology for identity authentication based on human facial feature information. Capture images or video streams containing human faces through cameras or cameras, and automatically detect and track human faces in the images, and then match and recognize detected faces. [0003] Face recognition has a wide range of applications, especially plays a very important role in many fields such as security and anti-terrorism, financial payment, access control attendance, identity recognition, etc. It involves domain knowledge such as biomedicine, pattern recognition, image processing, machine study etc. [0004] Face recognition algorithms mainly include: [0005] ①Template m...

AI Technical Summary

Problems solved by technology

The main problem with this method is that if the data set is large enough, it will take a lot of time to match and the accuracy will be reduced.

Overemphasizing the large distance between classes and ignoring the features with small distance between classes will eventually cause a large number of overlapping categories with small distance between classes, resulting in low final recognition accuracy

Images