Research On Face Recognition Algorithm Based On Improved LBP Operator

With the continuous development of computer science and technology,the status of face recognition technology has become more and more important.After decades of development,face recognition technology has made great progress.The recognition rate of the existing technology under ideal conditions has reached a very high level.However,due to too many uncontrollable factors in the human face,the existing technology is still not well applied in practice.In face recognition technology,the accuracy of recognition directly depends on the quality of the feature extraction algorithm adopted.Therefore,this paper mainly focuses on the face recognition of local binary pattern(LBP).LBP algorithm is a kind of texture description based on gray change.It has the advantages of simple calculation,easy popularization and strong classification ability,and it is widely used in the field of face recognition.However,LBP operator itself extracts and recognizes features in face recognition.There are still some limitations in order to solve this problem,this paper mainly completed the following research work:For the feature extraction algorithm combined with PCA and LBP,since PCA is an unsupervised algorithm,the algorithm combining PCA and LBP cannot extract effective image classification features.The 2DLDA algorithm is a direct image-based supervised feature extraction algorithm.It not only avoids the small sample problem,but also has fast calculation speed.Therefore,this paper proposes a feature extraction algorithm combining 2DLDA and LBP.The algorithm first blocks the image,uses LBP to encode each block and calculates the corresponding LBP histogram;then in order to use the 2DLDA algorithm,constructs the corresponding block LBP histogram matrix for each sample;finally use 2DLDA algorithm to find Projection matrix.In order to extract the facial feature information of face image from LBP operator,the shortcomings of the relationship between neighboring neighboring points are neglected.An improved LBP algorithm is proposed in this paper.The algorithm is first weighted according to the distance between the neighborhood point and the center pixel point.After computing,and then by increasing the size relationship between neighboring neighboring points,a new LBP operator value is obtained,and finally a facial feature histogram is obtained.The improved LBP algorithm described in the expression texture image will be more uniform and can better describe the face image.Based on the improved LBP algorithm,this paper proposes a feature extraction algorithm combining improved LBP and 2DLDA.This algorithm not only reduces the feature dimension effectively,but also obtains better recognition results.