Emotion recognition method based on facial features and voice features

The invention discloses an emotion recognition method based on facial features and voice features, and relates to the technical field of feature fusion, and the method comprises the steps: carrying out the framing of a collected audio/video containing a facial signal and a voice signal, and extracting a voice feature value and a facial feature value of each frame; constructing a K-means clustering number set according to the emotion types, and determining an optimal clustering number from the K-means clustering number set; the voice feature values and the face feature values are clustered into different clusters by using the optimal clustering number kbest, and the clusters containing both the face feature values and the voice feature values serve as fusion features; and inputting the fused features into an emotion recognition module, and outputting an emotion recognition result. According to the method, the k-means algorithm is introduced to fuse the expression features and the voice features, information of