Human Face Information Detection And Fusion Based On Computer Vision

The human face is an extremely important biological characteristics in the field of artificial intelligence research.The detection,analysis and processing of the face information is one of the prerequisites and key research directions for the realization of artificial intelligence of electronic equipment.There is a broad application prospects and market value,and it has become an important research focus.This paper is focused on the research of human face detection,the human eye location and analysis,facial expression feature extraction and dimension reduction,facial expression classification and expression recognition and other issues.On this basis,the facial information is for detection,fusion and application.The main works are as follows:1.In order to recognize the eyes state open or closed,the contour circle algorithm(CCA) is proposed.Firstly,the region of human face is detected based on the traditional Adaboost algorithm.In addition,the region of the eyes is traced and the pupils are positioned.Secondly,the pixels of the pupil region are removed to obtain the pure and better pixel set of the upper eyelid region according to the "grid" method.Then,the least squares method is utilized to fit the upper eyelid for the contour circle construction of the upper eyelid.Furthermore,the center position and radius of the contour circle are extracted as the feature vector.Finally,the eyes state open or closed is recognized according to the threshold criterion.This paper presents an eye blink recognition method based on binocular double threshold,which reduces the number of false detection.It is experimentally proved that CCA can accurately recognized the eye state open or closed,and the normal blinks or the fatigue blinks can also be distinguished.2.This paper is used to recognize the human emotion by the CLM(Constrained Local Model) method to get the shape feature model and the texture feature model from the manually calibrated face expression training set by means of the active appearance model.Then,the facial feature points are located and extracted,facial motion coding system divides the face region.Thereby forming a face motion unit.On the above basis,SVM(Support Vector Machine) is facial expressions and neutral expressions.And the eye state and expression recognition are fused to form a description of the human facial information.The algorithms and methods proposed in this paper are tested in the field of human emotion detection and artificial intelligence.The experimental results show that the contour circle algorithm can detect the eye state in real time,and the facial expression recognition based on the local constraint model and support vector machine method can identify the change of facial expression.To a certain extent,the intelligence of the machine is realized by the system.