Face alignment method

Abstract

The invention provides a face alignment method, which comprises the steps of S1, training, that is, a feature vector of each face feature point is extracted by using a face data set to serve as the input, the offset of feature vector is enabled to serve as the output, and a gradient regression tree is obtained through training; S2, tracking, that is, a face location box in an image is acquired byusing a face recognition algorithm, the face feature point is initialized, and the position of the initial feature point is updated by using the trained gradient regression tree so as to realize facealignment. The beneficial effect is that the face alignment method has the advantages of high detection accuracy, high environment adaptability and better generalization ability.

Description

technical field [0001] The invention relates to image processing, in particular to a face alignment method. Background technique [0002] With the rapid development of computer vision technology and image processing technology, face behavior analysis has been widely used in various fields, which has greatly promoted the development of human-computer interaction methods in various fields. At present, face recognition has been widely used in video conferencing, security protection, identity verification and other fields. [0003] The traditional face alignment method is generally to extract the feature description vector of the feature point, and use the description vector to train the classifier to realize the recognition of the feature point. This method has high requirements for positive and negative samples, low recognition accuracy, poor generalization ability, and is very prone to primary selection over-fitting problems. Contents of the invention [0004] In order to...

AI Technical Summary

Problems solved by technology

This method has high requirements for positive and negative samples, low recognition accuracy, poor generalization ability, and easy primary selection over-fitting problem

Images