Fast face recognition method in application environment of massive face database

Abstract

The present invention relates to a fast face recognition method in an application environment of a massive face database. The Bisecting K-Means (BKM) algorithm is combined with the PCA (principal component analysis) +LDA (linear discriminant analysis) algorithm to perform characteristic extraction from a large-scale face database. The method comprises the steps of: firstly, calculating a mean vector for each sample in the face database and then mapping the mean vector to a space with a specific dimension according to the principle that the intra-class aggregation is maximum, and the inter-class aggregation is minimum; then aggregating the mean vector in the dimension space; recombining the samples and grouped according to the cluster attribute after an aggregation result is obtained; and finally, extracting the characteristics on each cluster according to the linear subspace face characteristics extraction algorithm.

Description

technical field [0001] The invention is applied to the field of pattern recognition, and is specifically a method for fast face recognition under the application environment of massive face databases. Background technique [0002] Traditional linear subspace algorithms have certain limitations in massive face databases, such as: PCA+LDA based on Principal Component Analysis (PCA) and Linear Discriminant Analysis (LDA): PCA+LDA method The purpose of is to pursue an optimal projection plane, so that the projection of the original random vector on this projection plane has the characteristics of the smallest intra-class dispersion and the largest inter-class dispersion. [0003] Disadvantages of the traditional PCA+LDA technique: When the number of categories C of the original data is relatively small, it is relatively easy to find the best projection space of C-1 dimension. As the number of categories C increases, the dimensionality of its projection space is also increasing,...

AI Technical Summary

Problems solved by technology

[0003] Disadvantages of traditional PCA+LDA technology: When the number of categories C of the original data is relatively small, it is relatively easy to find the best projection space of C-1 dimension

As the number of categories C increases, the dimensionality of its projection space is also increasing, and the optimal projection space needs to meet more conditions, and such a space becomes more and more difficult to find

At the same time, due to the continuous increase in the number of categories, the inter-class distance of each category in the space is continuously shrinking, and the intra-class distance is relatively increasing relative to the inter-class distance, which brings difficulties to the design of the classifier and reduces the recognition rate. while the recognition time is increasing

Images