3D Vision Technology In Disordered Material Sorting

Material sorting is an important process of intelligent manufacturing/smart logistics.It is a key technology to use robots to realize automatic material sorting.The combination of vision technology and robot technology makes the robots replace the intelligent recognition function for the human eye,so that the robot can accurately and intelligently sort the materials,which can reduce the production cost of the industry and ensure the high product.In this paper,to combat the disordered sorting requirements of production line materials,we propose a technology of recognition and location combining 2D image SIFT features and 3D Depth images using machine vision to collect the disordered state images of materials.In this paper,aiming at the disordered sorting requirements of production line materials,using machine vision to collect the disordered state images of materials,combined with the characteristics of 2D images and 3D Depth images,the three-dimensional recognition and localization technology of NARF feature materials combining 2D image SIFT features and 3D Depth images is proposed.Firstly,we introduce the research background of the project,the research progress of stereo vision technology,and discuss the research content and work arrangement of this paper.Secondly,we design a 3D vision system for the process of disordered sorting material.To the beginning,the requirements of disordered sorting of materials are described.On this basis,the selection of hardware of the vision system is determined.The visual system is designed and the main modules are described.Thirdly,the preprocessing of the material image consists of two parts: 2D image preprocessing and preprocessing of 3D Depth image.In the 2D image preprocessing,the image is first enhanced,and then the image is filtered.Preferably,the image is segmented and the material area is extracted.In the depth image processing,the reason for the distortion of the 3D Depth image is introduced,and the method of removing the noise and improving the measurement accuracy is proposed.Fourthly,Obtaining Mark point images by using an area array camera and a 3D camera,respectively.The Mark points of the material are collected after the image preprocessing,and the conversion relationship between the 2D image coordinates and the3 D image coordinates are built;Then,a 2D image SIFT feature and a 3D Depth image NARF feature are extracted respectively,and the two features are merged.Finally,the2 D image and the 3D depth image are used.The mass centric coordinate and area coordinates for the material feature region in the three-dimensional Depth image are calculated by the mark point coordinate transformation relationship.Three-dimensional information of the material feature area is formed,and a three-dimensional model of the material feature area is generated to realize material identification and positioning.Finally,Experimental results and analysis are discussed in this work.Experiments show that the 3D vision technology is used to identify and locate the material sorting.The centroid coordinates of the material are calculated by the features,and the coordinate information of the feature points is sent to the the robot in the implementation of sorting.