Design And Implementation Of Facial Expression Recognition System Based On Sparse Optical Flow Method And HMM

Human emotion is a mysterious and complex intrinsic feature unique to human beings.It has always been regarded as one of the most essential differences between human and artificial intelligence.As one of the important elements of human emotion,facial expression helps to make up for the deficiency of artificial intelligence in understanding human emotions through the research of its automatic recognition.Therefore,in recent years,the research on facial expression recognition has become more and more important in computer vision.The difficulty of expression recognition is that there are differences between the facial features of different individuals.Each person expresses different expressions differently,and an expression may simultaneously combine the expressions of multiple emotions,sometimes even humans are difficult to complete.Correct understanding.All of the above factors bring great challenges to the automatic recognition of computers with expressions.This paper first focuses on the research and analysis of some of the more commonly used facial expression feature extraction and expression classification methods.Based on the difficulties and shortcomings encountered in these applications,the sparse optical flow method and hidden Markov model(HMM)are proposed solution.Based on this solution,a system for efficient facial expression classification for moving image sequences is designed and implemented.The optical flow method is suitable for extracting motion information in dynamic image sequences,but the automatic corner detection function does not necessarily select the appropriate point of interest.In this paper,the fusion of sparse optical flow method is proposed.The face-sparse optical flow feature extraction scheme of ASM method assists in selecting key feature points of expressions by ASM method,overcomes the deficiency of optical flow calculation for full-pixel pixels by dense optical flow method,and reduces the number of tracking optical flow points.This ensures that the feature extraction algorithm is more efficient and targeted.This paper proposes an automatic recognition system for facial expression based on optical flow feature detection and HMM model based on fusion ASM algorithm.Through the research,design and implementation of the system,it provides a real-time and efficient processing for automatic recognition of facial expressions.The program to some extent compensates for the shortcomings of artificial intelligence in understanding human emotions.The research work of this paper has certain theoretical research value,social value and application value.