Research on Facial Expression Recognition based on Motion Unit Combination Feature Matrix and Supporting Vector Machine

Abstract

Based on Haar and Adaboost methods, this paper uses genetic algorithm and cloud computing, collaborative simulation to improve facial expression recognition algorithm. It uses genetic algorithm to encode the movement element local feature combination, which improves marked effect of facial organslocal feature region. It uses cloud computing col- laborative simulation topology to establish facial local feature generalized matrix, which enhanced the calculation speed of the support vector machine expression classifier. In order to verify the efficiency and accuracy of the algorithm, this paper tests the facial expressions of the same individuals and different individuals using expression library. Throguh test- ing it is found that the improved method has higher facial expression recognition rate, faster computing speed and better performance. Throguh the analysis of results, the improved algorithm has higher facial expression recognition rate and it is higher in the same individual and different individuals, and the recognition rate of different individuals is the same as the average recognition rate, which verifies the reliability of the algorithm and provides a new method for the design of facial expression recognition algorithm.