Research Of 3D Face And Expression Recognition Based On Kinect

With the rapid development of modern technology,face recognition has been widely used in information security fields such as identity recognition and authentication login.The traditional 2D face recognition technology is susceptible to external interference,and its development is limited.But 3D face recognition due to the advantages richer facial features,has been more and more attention of researchers.By using the Kinect depth camera released by Microsoft,color and depth data can be obtained efficiently and conveniently,which can provide data support for 3D face recognition research.Therefore,this paper uses Kinect to obtain 3D face data,and research on 3D face and expression recognition.The main works can be summarized as follows.1、This paper applies Kinect camera to 3D face data acquisition,and completes the realtime acquisition of depth image and gray image through Kinect official SDK.Then,A algorithm based on Haar-Adaboost is used to detect the face in the gray image,and the position of the face region in the depth image is determined according to the coordinate relationship,and the face in the depth image is extracted.Finally,in order to remove the noise in the depth image,the face depth image is processed using bilateral filtering algorithm.The processed face image can be used for face and expression recognition,providing data support for later research.2、Based on the research of 3D face recognition algorithm,a 3D face recognition method based on 3DLBP and LBP dual modal fusion is proposed.The algorithm uses 3DLBP features to perform 3D face recognition on depth faces,and uses LBP features to perform 2D face recognition on gray faces.Finally,the recognition result of the decision fusion is obtained by weighted voting of the two modal results.The experimental results show that the proposed algorithm is robust to illumination and occlusion,and can achieve a recognition rate of 96.1% under the conditions of no illumination or occlusion,which indicates that it has a good recognition effect.3、For expression recognition,a 3D expression recognition method based on 3DLBP and HOG feature fusion is proposed.The method makes full use of the 3DLBP features that have been extracted in the face recognition,and introduces the HOG features to make up for the lack of 3DLBP features on the expression description.Finally,after dimension reduction and fusion of the two features,the SVM algorithm is used for classification and recognition.The experimental results show that the algorithm achieves an average recognition rate of 90.83% for the six basic expressions.4、Based on the above research,a 3D face recognition system based on expression encryption is designed and implemented.The expression recognition is added to the face authentication process,which realizes the double verification of the face and the expression.The system can solve the similar face and face counterfeiting problems well and has high security performance.The experimental results show that the system can achieve a low misrecognition rate under the condition of satisfying the high recognition rate.