Face recognition method based on multi-directional slgs feature description and performance cloud weighted fusion

Abstract

The invention discloses a face recognition method for multi-directional SLGS feature description and performance cloud weighted fusion, including the following steps: 1. Expand the existing SLGS algorithm from the direction angle to obtain the texture features of the face in different directions; 2. Based on the texture The characteristics adopt the method of hierarchical cross-processing to construct the base classifier, and form a performance cloud based on the recognition stability and reliability of the base classifier in different regions to obtain weights; 3. Through the weighted fusion of the base classifier, the target Face discriminative classification. The invention can use the multi-directional SLGS algorithm to fully describe the face image, and use the weight value of the base classifier obtained from the performance cloud to improve the recognition performance of the system and obtain a higher recognition rate.

Description

technical field [0001] The invention relates to a feature extraction method and integrated discrimination, and belongs to the field of pattern recognition, in particular to a multi-directional SLGS feature description and performance cloud weighted fusion face recognition method. Background technique [0002] Face recognition is a research hotspot in the field of image processing and computer vision in recent years. It has greatly promoted many related disciplines and has received extensive attention from researchers. Face recognition problems mainly develop along two main lines: feature description and object matching of face images. Feature description is the core step of face recognition. The ideal feature description should only reflect the changes in the essential attributes of the face due to the difference in appearance, and is insensitive to changes in expressions and lighting. Widely known feature extraction algorithms include PCA algorithm, Gabor algorithm, sparse...

AI Technical Summary

Problems solved by technology

But they generally have a defect: an important indicator for judging the performance of classifiers is that the model is statistically optimal for the training sample set, without considering the recognition stability and reliability of the base classifier in different regions of the sample space , that is, the specific situation of each sample, lacking a description of the reliability of a certain sample

Images