Positioning Theory Of The Three Tests Based On Machine Vision

Visual is the main way of human to percept the environment and understand the outside world. With electronic, optical, computer and other technologies are maturing, Machine vision technology have been rapid development. In recent years, advanced manufacturing, exploration and other measuring industries have been the rapid development, so the detection technology is supposed to have accurate, double-quick, automatic, intellectualized Features. Therefore, in this context, application of machine vision technology have also emerged a big leap. So, in recent years, it is active to research every part of the binocular vision technology, which detect and position objects by 3D information. The basic task of machine vision is to calculate capture three-dimensional geometric information of object using the two-dimensional image information from camera, and to reconstruct objects in three-dimensional space, or to identify objects.Papers develops a automatic double-quick accurate system of detecting and positioning space objects based on binocular vision machine theory, by researching how to choose Imaging Model, Camera Calibration, Image Processing. So that papers focus on the following areas.1,This paper research the geometric model of camera imaging in order to select a more ideal and realistic model. At present, the binocular vision has the parallel measurement model on the basis of binocular vision and the non-parallel measurement model on the basis of binocular vision. The former adapts to measure remote objects, and the other is suit for measuring the close objects. Non-parallel binocular vision measurement model is selected in the papers.2,Camera calibration, namely, determining of the various internal and external parameters. After researching the method and the related theory on the camera calibration, papers put forward a linear method of the camera calibration based on the coplanar points. By this method, we can create a new model of distortion calibration through step-by-step calibration, and linearly solve internal and external parameters completely only based on the coplanar points, without any movement of cameras. It not only avoid the tedious and instability nonlinear optimization, but also resolved camera calibration of the within parameters in other linear optimization and achieve calibrating simply and quickly with a high precision.3,Image preprocessing and feature extraction. There have been a lot of algorithms of image processing, which is perfect, so we can choose any algorithm being fit for, when we need them. But in binocular stereo vision, to find feature points and to match them is very difficult. After getting two-dimensional information of related points from feature matching, we can get three-dimensional information this point. At last we express three-dimensional space of object using the three-dimensional information of some marking point.