Study On Dual-View3D Measurement And Recognition System

Industrial inspection sometimes need to identify, locate of the workpiece. Sometimes there are more complex scenes in industrial inspection, there are different posture of objects and sometimes surface is covered, in this situation the method in2D image recognition is difficult to get good performance.3D recognition method have better adaptability in this situation.The traditional single-view3D laser measuring system hace obvious problems of missing data, and affected the accuracy of recognition system.This papper proposed dual-view3D laser identification system, using registration of two3D data obtained by single-view measuring system, more complete data is used to identify after registration, this kinds of dual-view3D recognition system have more high recognition accuracy.Research work of this paper is divided into the following areas:1.First set up the hardware platform of dual-view3D measuring system, do the research and design process to achieve a system involved image processing, stepper motor control, point cloud registration and other processing modules.2.By measuring the standard object, test the resolution and error of the system.3.Based on the basis of obtain accurate3D point cloud to do registration between the actual point cloud and the standard model, to complete the identification of the object system.4.By respectively using conventional single-view3D recognition systems and dual-view systems to test for different objects (metal,plaster,plastic,etc.), different scenarios (different inclination angles of objects, different blocking location). The results show that recognition accuracy has improved significantly when using dual-view recognition system.