Facial expression recognition method based on confidence region and multi-feature weighted fusion

Abstract

The invention discloses a facial expression recognition method based on a confidence region and multi-feature weighted fusion. The method comprises steps: 1, a face confidence region image and a face region image are acquired, wherein the face confidence region image at least comprises an eye and brow region and a mouth region; 2, the face confidence region image and the face region image are subjected to feature extraction to acquire corresponding initial features; 3, after dimension reduction and data normalization processing are carried out on the initial features, a fusion feature F is formed; 4, the fusion feature F serves as a classification recognition feature to be sent to a classifier for recognition; 5, training set feature data and test set feature data are selected, the training set feature data are inputted to a GRNN neural network for training, and corresponding training parameters are acquired; and 6, based on the training parameters, a density function is adopted to carry out prediction output on the test set feature data to acquire final classification recognition feature data. The method has the advantages of higher recognition efficiency and higher recognition accuracy.

Description

technical field [0001] The invention belongs to the field of computer vision, in particular to a facial expression recognition method based on confidence regions and multi-feature weighted fusion. Background technique [0002] Face recognition has been rising since the 1970s. It is one of the most popular research directions in the field of computer vision so far. It involves image processing, pattern recognition, computer vision, artificial intelligence, computer graphics, information theory, mathematics and statistics, Multidisciplinary interdisciplinary research topics of cutting-edge theories and algorithms in neuroscience, cognitive science, psychology and many other disciplines. Face recognition includes face detection and tracking, face verification, and various identification and other related technologies, which are widely used in intelligent video, intelligent robots, access control systems, and monitoring systems. Face recognition technology not only has importan...

AI Technical Summary

Problems solved by technology

[0006] (1) In terms of feature extraction, for static expression images, a single overall template matching method contains many irrelevant regional features, which affects the discriminative characteristics of the overall features, and it is difficult to achieve better recognition results

[0007] (2) Traditional classifiers are not effective for nonlinear data mapping, and the training time is long, and the recognition efficiency is not high

Images