Three-dimensional face occlusion discrimination method based on multi-bitmap tangent plane and multi-scale uLBP

Abstract

The invention provides a three-dimensional face occlusion discrimination method based on a multi-bitmap tangent plane and a multi-scale uLBP (equivalent mode LBP, uniform LBP), and the three-dimensional face occlusion discrimination method comprises the following steps: firstly inputting a three-dimensional face point cloud, and obtaining a depth map of the three-dimensional face point cloud; secondly, calculating a bitmap gray value of the depth map and obtaining eight bitmap tangent planes; extracting 6, 7 and 8 bitmap planes and synthesizing the 6, 7 and 8 bitmap planes to obtain a multi-bitmap tangent plane; according to x and y coordinates, carrying out 3 * 4 partitioning on a multi-bitmap tangent plane; extracting a multi-scale LBP feature descriptor in each sub-region and performingdimensionality reduction to obtain a uLBP feature descriptor; extracting each feature descriptor of 12 subareas of an input face, and connecting the feature descriptors in series to obtain a featurevector based on a multi-bitmap tangent plane and multi-scale uLBP; and training and testing in a support vector machine by adopting a Bosphorus database so as to complete judgment and identification of occlusion. The three-dimensional face occlusion discrimination method is high in detection and classification performance, simple in principle and easy to implement.

Description

technical field [0001] The invention belongs to the field of three-dimensional image recognition in computer vision, and in particular relates to a three-dimensional human face occlusion detection and classification method based on multi-bitmap tangent planes and multi-scale uLBP. Background technique [0002] The 3D face occlusion discrimination technology refers to the technology based on the 3D data of the human face, and the computer realizes the technology of occlusion discrimination of the human face. This technology has great application potential in research fields such as human-computer interaction and authentication. Compared with two-dimensional data, three-dimensional face data is not affected by factors such as light, posture, angle, etc., and contains more geometric information and topological features. Therefore, face recognition research based on three-dimensional face data has gained a lot of attention in recent years. wider attention. Facial changes due t...

AI Technical Summary

Problems solved by technology

The method based on facial curves or radial curves, whose feature structure depends on the position of the feature points of the facial features, will have different effects on different types of occlusion due to the different selection methods of the curves; the multi-modal method will not only point cloud Color, brightness and other information can be applied, but it is also limited by these other information, and additional RGB images of face point clouds need to be collected; methods based on point cloud geometric information can describe the geometric features of face deformation, such features include method Vector, curvature and shape index, etc., but the amount of data is large

Images