Face recognition method

Abstract

The invention discloses a face recognition method. The method includes: based on a Huffman tree and an Euclidean distance, establishing a face feature group of a Huffman tree structure established bystarting from a root node face image; when face recognition is needed, obtaining a face image pic1 needing to be identified; calculating the Euclidean distance between the characteristic value feature1 of the pic1 and each node of the Huffman tree; selecting a node closest to a pic1 Euclidean distance for matching, directly and accurately searching out the information of the same face; when a newface image is added, only matching the face with a root node face, adding the face into a nearest face group, and once the face image is added, re-creating the Huffman tree and selecting the root nodeface. Therefore, the efficiency and accuracy of face recognition are improved, and the usability is higher. The face recognition method has the advantages of being high in face recognition efficiencyand precision and high in usability.

Description

technical field [0001] The invention relates to a face recognition method, in particular to a face recognition method with high precision and good timeliness. Background technique [0002] With the continuous deepening of artificial intelligence technology in the field of smart community business, more and more artificial intelligence technologies are implemented in specific applications of specific smart communities. Among them, image processing technology is the most widely used. Through image processing, especially face recognition technology, it is most suitable for specific business fields such as community time alarm and authorization authentication, and is developing rapidly. However, although the recognition degree of face recognition technology is getting higher and higher, the complexity of the corresponding algorithm is also increasing, and the performance requirements of the machine are also getting higher and higher, especially in some application scenarios with...

AI Technical Summary

Problems solved by technology

However, although the recognition degree of face recognition technology is getting higher and higher, the complexity of the corresponding algorithm is also increasing, and the performance requirements of the machine are also getting higher and higher, especially in some application scenarios with high real-time requirements, which not only require Accuracy to be usable, but also speed to meet timeliness

However, the amount of calculation and timeliness related to performance are always trade-offs.

Especially when functions such as image search by image and image search by image are implemented, if all face images with the same face and corresponding video files that have appeared in the video are found from tens of thousands of faces, it is often It takes at least four or five seconds, which is fatal to the usability requirements of the product

[0003] At present, most of the processing methods use image preprocessing, calculate face features in advance, and then perform digital comparison of face features, which can solve part of the efficiency of this problem problem, but the accuracy is difficult to improve, and some clustering algorithms are used for preprocessing, but the effect is not very good

Images