Research On Face Expression Recognition Algorithm Based On Improved Gabor Wavelet Feature Extraction

With the continuous development of hot areas such as artificial intelligence,computer vision,as an important part of intelligent human-computer interaction,facial expression recognition has become the hot topics in the study of many scholars at home and abroad.This paper mainly studies the facial expression recognition algorithm of static image.Firstly,according to the gray information of the eye,the image processing method is used to locate the human eye position quickly and accurately.According to the human eye position and the facial structure,the effective expression area is divided and normalized,and the standard pure expression image is obtained.Secondly,in order to overcome the problem that the feature information extracted by Gabor wavelet is not comprehensive enough and the dimension disaster,the improved method is presented.In this paper,two methods of Gabor wavelet and Asymmetric Local Gradient Code are analyzed.This paper combines their respective advantages and the extracted Gabor feature is encoded by AR-LGC operator to obtain the AR-LGGC composite feature of the image.The coded Gabor feature maps are divided,and histogram statistics are performed for each sub-block.Then the histogram of each sub-block is concatenated to form the facial expression vector.The algorithm not only can get more detailed texture gradient information,but also can analyze the image in multiple scales and multiple directions.Thirdly,in order to overcome the problem of dimensionality disaster caused by multi-scale and multi-directional Gabor feature extraction,the Gabor features of the same scale in different directions are fused and encoded under the inspiration of Local Binary Pattern.The algorithm greatly reduces the feature dimension,while not losing important decision information.Finally,SVM is used for expression classification and recognition.In this paper,the kernel parameter ? and penalty factor C of SVM are optimized by using grid search and K-fold cross validation method,which improves the generalization ability and classification accuracy of SVM classifier.In the Matlab platform,the improved algorithm is simulated.The validity of the algorithm is proved,and a facial expression recognition system is designed.