A Face Recognition Method Based on Local Multivalued Patterns Based on Weber's Law

Abstract

The invention discloses a local multi-value pattern face recognition method based on the Weber law. The method includes a training stage and a recognition stage. In the training stage, firstly features of training samples are extracted, and three-level encoding is performed on the samples; then three-level encoding is converted into a Uniform mode, even block processing without overlapping is performed on images on which three-level encoding is performed, feature histograms of blocks are extracted, and all levels of the feature histograms of all the blocks are connected to obtain an overall feature histogram of a face image. In the recognition state, feature histograms of samples to be detected are extracted through the feature extraction method to be used as features to be detected, and then x2 probability statistics and nearest-neighbor classification methods are used for recognizing the samples to be detected. In face recognition, the method has better robustness in face illumination and posture, and the face recognition rate is improved.

Description

technical field [0001] The present invention relates to technologies such as pattern recognition, image processing and computer vision, and in particular to a face recognition method based on Weber-based Local Multiple Patterns (WLMP). Background technique [0002] In recent years, face recognition technology has been widely used in government, military, banking, social welfare, e-commerce, security defense and other fields. Compared with traditional identification methods, the biggest feature of face recognition is that it is more secure and confidential and convenience. The unique activity discrimination ability of face recognition ensures that others cannot deceive the recognition system with inactive photos or even human heads. In addition, face recognition is fast and difficult to detect. Compared with other biometric technologies, face recognition is an automatic recognition technology that can recognize several times in one second. The unobtrusive feature is also i...

AI Technical Summary

Problems solved by technology

When Yue Ming (see Reference 5) et al. used LBP for 3D face recognition based on depth maps, they found that only binarization of adjacent and central samples could not effectively distinguish the difference in depth. For this reason, they proposed 3DLBP. The 3DLBP operator re-encodes the difference between the center and adjacent pixels, but 3DLBP cannot automatically determine the upper limit, and the choice of the upper limit has a great impact on the recognition result

Images