Face recognition method based on color images

Abstract

The invention discloses a face recognition method based on color images. The method effectively uses color information of color images to convert the color images into single-color images on basis of CSDA (correlation similarity discriminant analysis) criterion, and is optimal as compared with the subsequent CSDA recognition method in terms of obtained color faces. The method includes: obtaining a feature space by the aid of a training image sample through similar features in a training stage; using projection factors of the training image sample in the feature space as features of the training image sample; projecting a test image sample to the feature space in a test stage; using factors obtained after the test image sample is projected to the feature space as features of the test image sample; matching the features of the training image sample with the features of the test image sample by a classification method to obtain classification results; using a similar feature extraction model for converting color images in concentrated training into single-color images according to an optimized combination coefficient vector X*; and obtaining features of training images according to an optimized projection matrix W*.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a color image-based face recognition method. Background technique [0002] Nowadays, face recognition has become a research hotspot in the field of pattern recognition and artificial intelligence. However, most of the current face recognition algorithms are based on grayscale images. Recent studies have shown that color images can provide useful information for face recognition. face recognition rate. There is a low-dimensional nonlinear manifold structure in the face space, and the geometric structure of the manifold is embedded in the similarity of samples. Such as figure 1 As shown, Zhang Taiping proposed a correlation-based similar feature extraction method called Correlation Similarity Discriminant Analysis (CSDA), which can effectively extract the manifold structure hidden between sample similarities. However, the CSDA method is proposed based on gr...

AI Technical Summary

Problems solved by technology

However, the CSDA method is proposed based on grayscale images. This conversion actually loses the color information of the face image.

Images