Spatial Surface Reconstruction Method Based On Monocular Vision And Active Laser Tag

Computer vision is a science that studies how to make the computer realize the function of "seeing". It is not only able to simulate the function of human eye, but also capable of completing the job beyond the latter’s ability. As an emerging discipline, it is an important research area full of challenges in fields of both engineering and science and has been drawing increasing attention from scholars. Computer vision research, ranging from physics, mathematics, optics, computer science, signal processing science, biology, neurophysiology to cognitive science, is an integrated frontier science with extensive application prospect. This thesis, in order to obtain a space object position through combining monocular vision and laser ring, implements a series of exploration and study.Through analyzing the present research situation of monocular vision at home and abroad, this thesis clarifies the research direction in this study and builds models to analyze the issues needing to be resolved such as edge extraction method, curve fitting method, solution of the top position in conical laser ring cone vertex and solution of a space object position in accordance with the image information. Because it is hard to capture a cylindrical ring, this thesis suggests adopting conical laser ring to conduct an analysis with monocular vision.As elliptic spot is formed after conical laser ring irradiates an arbitrary space plane, least square ellipse fitting method is put forward to obtain elliptic equations of the CCD surface spot; however, after laser ring irradiates an arbitrary space curve, no elliptic spot can be formed. In this condition, spot edge curve will be divided into the upper and lower parts, and again least squares polynomial method will be used for fitting the upper and lower spots to attain spot polynomial.It takes three steps to determine the top position of conical laser ring:first of all, two circular plane parallel to the CCD surface is calibrated, center line of the two circular plane is paralleled to the camera optical axis, laser position is adjusted to make sure all the light go through the circular plane center at the same time, and then coordinates of the circular plane center are secured to determine the coordinates of the top position of conical laser ring in two directions; secondly, laser ring irradiates the two circular plane respectively and ellipse fitting is adopted to obtain the imaging facula center on the CCD surface; lastly, spatial linear equation is applied to determine another coordinate of the top position in conical laser ring.In order to obtain a space object position, firstly, a space plane parallel to the CCD surface needs to be calibrated, and its spot edge data can be calculated by applying camera imaging and space line equation. Then, a point M at the spot edge is selected to cooperate with the camera focus and the three points on the top position of conical laser ring to construct a plane which in turn intersects the imaging facula on the CCD surface, resulting in a feature point. This feature point is connected with the camera focus to construct one straight line; the top position of conical laser ring is connected with the point M to construct the other straight line. The intersection point of the two straight lines is the very point in the space object. That is to say, the solution of the former equals that of the latter. If points along the spot edge are selected and analyzed with the movement of laser ring, location of every point in the space object can be obtained. If the object is an arbitrary space plane, least squares plane fitting method can be adopted to obtain the position of this plane in accordance with location of every point in the space object; however, if the object is an arbitrary space curve, curve construction method can be applied to get the curve shape.The experimental platform is built to attain the focus and the location of image coordinate system relative to the world coordinate system according to the calibrated internal and external camera parameters, and to analyze the error causes based on the experimental results.