The Study On Computer Stereo Vision And Its Application In Measureing Stone

Computer Vision is an important branch of Artificial Intelligence,which would simulate human's sense behaviors and thoughts. Thistechnique analysis 2D image from the image sensor, and comprehendthe complex 3D scene. Under the background of the research about thisstone measuring system, this paper discusses the related theory andtechnique about the computer vision deeply.As an important branch of modern mathematics, Group Theory is anexcellent tool to research symmetry and transform. As we know, thereare lots of transform in computer vision, so we involved Group Theoryto deal with these problems. This paper put forward a new cameracalibration method based on Group Theory, and also discusses GroupTheory in describe the properties of the object and even positionbetween multi-objects. Camera demarcate is a key problem in computer vision, whichpurpose is to know the parameters of the camera model, then peoplecan resolve the relations between the 2D image and the 3D object.Based on the simple but exact pin-hole model, we analysis the threereference frame. No matter the reference frame or the CCD camerarotate and shift, the matrix could be defined a Group per multiplication,pure rotation Group R, pure shift Group T. And it shows threesub-group of the Group R, Group T. In this paper, we put forward anew concept. People usually concern some objects or districts, wename these districts objects with coordinate, color, and moreinformation. These objects should be expanded, so that we get theGroup GW,GC,GG. The compound relation between the Group GWand Group GG is the parameter we want to know through Camerademarcate. Other thesis have confirmed it could be resolved by BPneural net.Also this paper analysis single character of single object ( rotation ofa screw cap), combined character of multi objects with Group Theory.The results predigest the expression of the problem, and improve theefficiency to resolve the problem.The stone machining expert system will realize rapid measure andmachining. To complete the task, there are two kinds of ways. Wecompared these typical ways and per the target choose the wholeproject: PC based Stereo Vision measuring system. To build asuccessful system, we must consider the size of the image, the speed,the precision (Includingother factors, temperature, humidity, dust andso on). The author refers a flow to build a vision measuring system,including CCD camera, lamp-house, camera lens, image card, andsoftware system.Stereo Vision studies on the technique by which one can deduce the3-D information of an object (such as shape, range, and so on) from astereoscopic image pair. Especially, the stereo matching is the difficultand important part,including matching elements, matching rules andthe arithmetic. Here, two kind of stereo matching approaches arestudied. How to choose the point and line and other matching elements?How to choose matching rules? All these must be decided per thedifferent situation.In other chapters, this paper also puts forward the concept of imagesegmentation, image smoothing, and morphologic compute. Firstly, thispaper puts forward the concept of image segmentation. Study andanalyze ordinary methods about these, which includes edge detection,edge joining and soon.Despite we discuss some problem in computer vision field, such asapply Group Theory in Camera calibration, it is not enough in deepnessand in extent. There is still many problem needed to be researched andsolved. For all that, we believe that our work in theory and practice willbe useful for latter research because of the broadly practicallysignification of computer vision.