Acquisition and Processing of Depth Image Based on Kinect
Download as PDF
Jinrong Tian, Ying Sun, and Chao Xu
The development of science and technology puts forward higher requirements for manufacturing industry. In order to meet the functional requirements, complex surface can be designed and manufactured. However, the processing of complex surface has the problems of difficulties in mechanical processing and low operation efficiency and low precision in manual. In recent years, machine vision has developed rapidly, and one of its branches, three-dimensional reconstruction technology, has attracted more and more attention. The acquisition of feature parameters from 3D reconstruction of complex surface and simulation and analysis are the foundation and premise of subsequent processing. The application of machine vision technology in mechanical processing is the inevitable trend of the development of automation and intelligence in manufacturing industry. Therefore, this paper proposes a Kinect-based method to obtain the depth image, preprocess and transform the depth image to get the three-dimensional point cloud of the workpiece surface. ICP algorithm is used to register the point cloud, obtaining the three-dimensional coordinates of the surface space, and realizing the three-dimensional reconstruction of the surface.
Curved surface, Depth image, Point cloud, Three-dimensional reconstruction