Color face recognition method based on multi-kernel dictionary learning

Abstract

The invention discloses a color face recognition method based on multi-core dictionary learning. The multi-core learning technology is applied to the dictionary learning and sparse coding process of color face data, and the color face image training samples are trained by designing kernel function selection criteria. Select the optimal kernel mapping function for the three color components of the set, and then learn three feature extraction coefficient matrices, structured dictionaries and corresponding sparse coding for the three color components after kernel mapping, and then use the learned dictionary to treat Non-linear features of identified samples are sparsely encoded, and classified and identified based on reconstruction errors. The invention has higher recognition effect and better robustness to image quality problems.

Description

technical field [0001] The invention specifically relates to a color face recognition method based on multi-core dictionary learning, and belongs to the technical field of face recognition. Background technique [0002] Existing color face recognition methods based on multi-core learning technology include: [0003] For a color face image training sample set X, let n represent the number of all color face image training samples, X R ∈ R d×n 、X G ∈ R d×n 、X B ∈ R d×n respectively represent the three color component sample sets of R, G, and B, and d represents the dimension of the color component sample; φ R :R d →H R , φ G :R d →H G , φ B :R d →H B Represents three kernel maps, which respectively transform the three color component samples of R, G, and B from the original d-dimensional linear space R d Mapped to three nonlinear high-dimensional kernel spaces H R 、H G 、H B , H i The dimension of k i Denotes the kernel map φ i The corresponding kernel fun...

AI Technical Summary

Problems solved by technology

[0013] The CD-MK-DA method and the CD-MK-DCA method use three different nonlinear kernel maps for the three color components respectively, but they do not consider how to choose an appropriate kernel map function; and these two methods do not consider how to deal with Image quality problems such as noise interference, partial occlusion and image corrosion, the recognition effect is greatly affected in the color face recognition task with these problems

Images