Research On Local Feature Of Three Dimensional Face Recognition

Since the introduction of the two-dimensional face recognition technology in the last century,it has achieved satisfactory results so far,and the two-dimensional face recognition technology has been applied to various fields,but it is still subject to changes in face posture and environmental illumination.The impact of etc.The rapid development of 3D face recognition technology is to solve the above bottleneck problem of 2D face recognition.3D face recognition is not affected by changes in face pose and ambient illumination but is susceptible to changes in expression,and each 3D face cloud contains tens of thousands of points,and the calculation speed is greatly affected.Therefore,the purpose of this paper is to improve the expression of 3D point cloud face,reduce the number of points participating in the calculation,and improve the speed of the algorithm;another purpose is to improve the robustness of 3D facial expression changes.The main experience of the 3D face recognition algorithm has been divided into two major stages: one is to match the three-dimensional face with the overall feature in the early days.This method is computationally intensive and can only be operated on a complete human face.It is sensitive to face data missing and occlusion;the other stage is 3D face recognition algorithm based on local features,which is also the mainstream direction of current research.These algorithms are more flexible and more robust to occlusion and missing.The main research methods and results of this paper mainly include the following aspects.In the preprocessing stage,according to the special position and geometrical properties of the nose tip,a nose-point detection algorithm based on the support vector machine is proposed,and the detection efficiency is good.Aiming at the excessive amount of 3D face point cloud data,a method of simplifying 3D face point cloud using the geodesic contour and radial curve based on the nose point is proposed,and then the shape analysis algorithm is introduced to simplify the face calculation.The geodesic distance of the corresponding three-dimensional curve is used as the basis for classification.The nearest neighbor algorithm is used.The experimental results show that the algorithm has a high recognition rate.The deformation caused by the change of expression of 3D face affects the recognition effect.It is proposed to divide the 3D face according to the characteristics of its semirigid and non-rigid regions and subdivide the 3D face into 11 blocks,using the covariance matrix.Representing the local features of each block,the sparse expression classifier is improved,and the weighted multi-task sparse representation classifier is proposed.The experimental results show that the algorithm has a higher recognition rate and expression change robustness.