Emotion recognition method based on multilevel decision fusion of fuzzy rule

The invention belongs to the field of mode recognition and emotion recognition, and discloses a multilevel decision fusion emotion recognition method based on a fuzzy rule, which comprises the following steps of: 1, acquiring an audio-visual emotion data set, performing voice MFCC (Mel Frequency Cepstrum Coefficient) feature extraction in a voice mode, and performing facial expression image preprocessing in a visual mode; 2, a visual facial expression recognition model is established, the input of the model is a facial expression image, and the output of the model is an emotion classification result; sentiment classification results comprise anger, aversion, afraid, happiness, neutrality, sadness and surprise; 3, establishing a decision-level fusion speech emotion recognition model based on a fuzzy rule; and step 4, establishing a vision and voice bimodal emotion recognition model based on a fuzzy rule. According to the method, the modal weight is distributed according to the contribution of the modal through