With the development of information and identity verification technology,face recognition has become a research hotspot in the field of computer vision because of its convenience and friendliness.At present,the face recognition technology has achieved good performance under the ideal condition.However,there are many uncontrollable factors in practical applications,such as light influence,attitude change,face occlusion and other problems.The attitude change of face is the most common problem,it will distort the face feature and affect detection and recognition.So this paper makes a study on multi-view face recognition.This paper studies multi-view face detection.A new multi-view face detection algorithm based on skin color and improved Ada Boost is proposed.The algorithm uses skin color as the rough detection,which can determine potential face area.We analyze the process and shortcomings of Ada Boost algorithm and propose new multi-view face detection algorithm.We divide the face pose change into five intervals,train Ada Boost detector separately,and form a multi-view detector.The experimental results show that the algorithm can locate the multi-view faces between [-90°,90°].In this paper,a new multi-view face recognition algorithm based on attitude estimation is proposed.The attitude estimation algorithm based on improved ASM algorithm which make the edge location more accurate.The T structure is made up of three features of the eyes and nose,and the estimation of the attitude is carried out according to the mapping relationship between the rotation angle and the face distance.The face training library is divided into three intervals according to the attitude change.Face feature extraction is based on PCA+LDA algorithm,PCA algorithm is used to reduce dimension,LDA algorithm is projected to classify.Finally,according to the distance function,multi-view face recognition between [-60°,60°] is realized.The multi-view face recognition system uses vs2013 as the development environment and uses the Open CV library to realize the multi-view face detection and recognition function.Through the test of CAS-PEAL multi-view face database,the recognition rate above 70% is achieved. |