Feature extraction method for face recognition based on Weber local symmetric graph structure

Abstract

The invention relates to a feature extraction algorithm for face recognition based on Weber's local symmetric graph structure. Construct the graph structure above; calculate the difference sum Xg between adjacent pixels in the graph structure; calculate the average value Xm of the pixels involved in the graph structure in the neighborhood; calculate the eigenvalue of the central pixel in the neighborhood. The present invention introduces Weber's law, fuses the obtained direction and differential excitation information, can accurately describe the features of the face and finally obtain the feature value of the image, and describes the feature information of the face image more comprehensively. In the database, the face features can be well described, and the face recognition rate is improved when the image processing time is shortened, and the generalization ability is good, which can be widely used in the field of face recognition.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a facial recognition feature extraction algorithm (WSLGS) based on a Weber local symmetric graph structure. Background technique [0002] Feature extraction algorithm is a very important step in face recognition system. A good feature extraction algorithm can improve the speed and accuracy of face recognition. Feature extraction algorithms can generally be divided into two types based on global features and based on local features. Algorithms based on global features are suitable for describing the entire image, but the problem is that the foreground and background cannot be distinguished, especially if the region of interest is occluded, the global features will be destroyed. The algorithm based on local features solves this problem well, and it can restore important information through some unoccluded feature points when the object is disturbed. [0003] At present,...

AI Technical Summary

Problems solved by technology

Algorithms based on global features are suitable for describing the entire image, but the problem is that the foreground and background cannot be distinguished, especially if the area of ​​interest is occluded, the global features will be destroyed

The disadvantage of this algorithm is that during the whole calculation process, the eigenvalue of the central pixel is converted between binary and decimal, which to some extent causes the loss of information

[0005] The neighborhood ranges selected by the above two algorithms are 3×3 and 3×4 respectively. Under certain conditions, some overall features of the face image will be segmented, resulting in recognition errors.

Images