An optimized processing method for 3D face cloud point cloud based on kinect camera

Abstract

The invention belongs to the technical field of three-dimensional reconstruction, and relates to a method for optimizing and processing three-dimensional point clouds of human faces based on a kinect camera. The technical solution of the present invention is different from the traditional processing of three-dimensional point clouds. The solution of the present invention uses the four-channel image RGB-D of the Kinect camera to directly process the depth information to obtain the point cloud optimization effect. The method of the present invention utilizes the existing Kinect depth camera to optimize the processing flow of the face three-dimensional point cloud, avoiding the expensive price of traditional high-precision equipment, or the problem of low precision of general algorithms; in the depth optimization step, using When the obtained optimized depth is used for surface restoration, the normal vector is not used for direct fusion, because the fusion of the normal vector to solve the surface will produce offset errors. Here, the geometric relationship between the depth gradient and the normal vector is used to solve the problem, and the final result is obtained directly. The depth, and finally use the change of coordinates to get a high-quality 3D point cloud.

Description

technical field [0001] The invention belongs to the technical field of three-dimensional reconstruction, and relates to a method for optimizing and processing three-dimensional point clouds of human faces based on a kinect camera. Background technique [0002] At present, 3D reconstruction technology is a research hotspot in the field of computer vision, and the 3D model of human face is a relatively important direction, which is often used in film animation production, games, medical treatment and other related scenes. Currently, there are three methods for acquiring 3D faces: manual modeling, instrumental acquisition, and image-based modeling. As the earliest 3D modeling method, manual modeling requires professionals to carry out; in instrument acquisition, based on structured light and laser scanners are typical representatives, but generally speaking, the cost of instruments is relatively high; image-based modeling The main method is to use a certain algorithm to obtain...

AI Technical Summary

Problems solved by technology

[0004] What the present invention intends to solve is to propose a method for optimizing the processing of 3D point cloud of human face not based on kinect camera in view of the problems in the above-mentioned traditional methods

Images