Enhanced recognition method for facial expression image

Abstract

The invention discloses an enhanced recognition method for a facial expression image, and the method comprises the following steps: 1), carrying out the face positioning through an Adaboost cascade detector based on Haar features, framing a face part, carrying out the cutting, carrying out the image preprocessing of a cut image, and storing the image; 2) establishing a mapping relationship betweenthe human face appearance and the human face shape by using a cascade regression tree algorithm in the regression model, and extracting facial feature points; 3) calculating a corresponding Euclideandistance by using the facial expression representation model to obtain a six-element array representing facial expression features; and 4) training a classification model by using a random forest algorithm, and inputting the six-element array into the trained model to realize classification and identification. According to the invention, the facial expression image with certain head deflection can be well recognized on the basis of recognizing the standard posture facial expression image, the recognition efficiency is high, the operation speed is high, the actual application requirements aremet, and the method is more suitable for actual scenes.

Description

technical field [0001] The invention relates to the technical fields of computer vision and pattern recognition, in particular to a method for enhancing recognition of facial expression images. Background technique [0002] Facial expression recognition technology is to analyze the specific mood of a person by extracting a specific expression image in a picture or video, in order to perform better human-computer interaction. In view of its high degree of information, facial expression recognition plays an important role in psychological analysis, clinical medicine, safe driving, and criminal investigation. The facial expression recognition at the present stage is mainly aimed at the standard posture, that is, the frontal facial expression. However, under normal circumstances, people tend to unconsciously produce a certain head deflection when making expressions, and in practical applications, it may Faced with many complex situations, its recognition effect often cannot mee...

AI Technical Summary

Problems solved by technology

The method based on the key points of the face mainly relies on the geometric model to locate the key points and then recognize them, which requires a large number of samples to mark the key points, and it is difficult to automatically mark the key points (He Jun, He Zhongwen, Cai Jianfeng, Fang Lingzhi. A A new face expression recognition method with deflection angle [J]. Computer Applied Research, 2018, 35(01): 282-286.); Appearance-based method is to obtain the local or global expression features of the face in different poses, and reduce the image The interference of factors that have nothing to do with expression avoids the problem of difficult key point extraction, but the recognition effect is general (Wang Chenxing, Liang Yu. A new expression recognition method [J]. Electronic Technology and Software Engineering, 2018 (06): 67.); Posture-based methods can be divided into two types, one is to group the expression library according to different poses of faces, and train and classify them in groups; the other is to establish the relationship between non-frontal faces and frontal face samples, Map non-frontal faces to frontal faces, and then classify and recognize frontal faces (Zheng Wenming, Feng Tiancong. Non-frontal facial expression recognition method based on pose normalization [P]. Jiangsu: CN103400105A, 2013- 11-20.), this method works well, but due to the complexity of the algorithm and slow operation, it is not suitable for practical application

Images