| The research purpose of computer vision is to make computer perceive surroundings information. This not only make computer perceive geometry information, including shape, position, gesture and movement, but also can describe, storage, identify and understand it. As an independent subject, computer vision is recognized widely, it has been applied to remote sensing image, character identifying, medical image processing and military affairs etc. At present, the experts are devoting to apply computer vision to the auto crashworthy system. They try to obtain the distance between two cars through processing the photo of cars, so that the driver could brake in time. Although the investigation of this is seldom, with the development of the computer vision theory and the image processing technology, it has comprehensive research foreground. This thesis try to syncretize the theory of computer vision, arithmetic and image processing, the camcorder can achieve a two-dimensional image of an object, the image will be analyzed and processed, so the three-dimensional position of the object can be restore. This thesis presents a distance measure system based on new method for determining the three-dimensional position and orientation of an object. At present, the investigation of distance measure through the image is seldom, so the distance measure system based on the method for determining the three-dimensional position and orientation of an object is very significant, meanwhile, it is important aspect of the computer vision application It is obvious that this thesis mainly consist of three parts, the camcorder calibration, module of image processing and the reconstruction of the space dot. We discuss as follows.1. The thesis analyzes linearity model (pin-hole model ) and its calibration, bring forwards a new calibration method, namely the least two multiplication. Image commutation involves the switch between different coordinate system. If the three-dimensional object projects the two-dimensional image, involved coordinate frame consist of the world coordinate frame, the image coordinate frame and the camcorder coordinate frame. The relation between three-dimensional dot and two-dimensional image dot is determined by the project matrix. The project matrix is 3*4 matrix. The process determining every element of the matrix is camcorder calibration. This method use two camcorders named and , the light axis of and image plane intersect the dot , the light axis of and image plane intersect the dot , the midpoint of the line is the origin of the world coordinate frame. The projective matrix of two camcorders is matrix M1 and matrix M2. If there are black grids on the reference object , it is easy to calibrate camcorder. Each black grid has four acmes (altogether 48 dots), their three-dimensional coordinate are known accurately, namely () . Every acmes corresponds a two-dimensional dot on the image obtained by , image dot is labeled . Similarly, image dot obtained by is labeled . We can calculate matrix and through above-mentioned dot. The project matrix contains twelve unknown element, if we can know six world coordinate and their image coordinate, we can calculate project matrix. When we calculate the matrix, we use forty-eight dots. We also can use more dots, namely the quantity of the formula greatly exceeds the quantity of the unknown element, and then the precision of the answer will greatly increase. During calibration, it can reconstruct the object through the projective matrix, avoid solving the parameter of the camera inside and outside, decrease calculation complexity.2. The module of image processing The thesis begins the representative image processing, narrates the method of image enhancement (including using the equilibrium algorithm, contrast enhancement etc), outstanding edge (including differential coefficient operator, log operator, Canny operator), distill beeline using Hough transform, analyzes and compares every means, forms the module of image...
|
PDF Full Text Request