Based on gb(2d) 2 Face recognition method of pcanet deep convolution model

Abstract

The invention discloses a GB(2D) based 2 Face recognition method for PCANet deep convolutional model. The model training method includes the following steps: sending preprocessed face samples to the first feature extraction layer in turn, scanning multiple sub-blocks from the acquired Gabor feature image and removing the mean value, using (2D) 2 PCA extracts the optimal projection axis, and convolves with the original sample of the training set to obtain the first layer feature map; send the first layer feature map to the second feature extraction layer, repeat the previous steps to obtain the second layer feature map; The output feature map is optimized, and the local area histogram is calculated and stitched as the final feature; the final feature is sent to the linear SVM classifier to obtain an optimized face recognition model. The invention can automatically learn effective feature expression, not only has good locality, but also has good robustness to illumination, expression and noise, etc., and improves the recognition performance of face identity.

Description

[0001] Technical field: [0002] The invention belongs to the field of machine vision, in particular to a GB(2D) based 2 Face recognition method for PCANet deep convolutional model. [0003] Background technique: [0004] Face recognition technology is a technology that uses computers to analyze face videos or images, extract facial features from them, and identify identities through these features. [0005] At present, face recognition technology is developing rapidly, and a large number of research results have been obtained. Common face recognition algorithms can be divided into several categories: face recognition based on geometric features, face recognition based on subspace analysis, face recognition based on elastic matching, and face recognition based on hidden Markov model. Identity recognition, face recognition based on neural network and face recognition based on 3D. For example, Takatsugu et al. [1] used an elastic matching method based on dynamic link structure...

AI Technical Summary

Problems solved by technology

[0006] The present invention aims at the problems existing in the above-mentioned face recognition, and proposes a method based on GB(2D) 2 The face recognition method of the PCANet deep convolution model not only absorbs the advantages of the depth model and Gabor filtering, but also can extract more abstract features in the data, and is robust to factors such as illumination, expression, and occlusion, and overcomes the convolution Disadvantages of time-consuming neural network and large number of labels

Images