| Recently, with the high popularization of computer equipment and rapid development of its software and hardware, the computer has become an integral part of human’s daily life. With the development process of industrial automation and rapid advance of computer calculation speed and image acquisition, computer vision theory has gradually become the computer field research hot spot.Binocular stereo vision technology is an important branch in the field of computer vision areas. With the help of imaging theory of the human eye, using image acquisition equipment acquisition image as input, through the stereo matching and3D reconstructing the position information of the original object, binocular stereo vision technology can realize the function of computer simulation of the eyes. The specific implementation process as follows:first of all, using two cameras on the public scene for image acquisition in different positions, and then through calculating the corresponding points of the parallax of space point in the image, at last, restoring Three Dimensional Coordinate of this space point. Nowadays, binocular stereo vision technology has been widely used in computer graphics, cognitive psychology, artificial intelligence, and many other fields, and it has successfully provided preferable solutions to many challenging problem in aforesaid fields many times. In addition, binocular stereo vision technology has also been widespread applied to our production and life, for instance, robot navigation system, parts orientation and detection, and so forth. This article is based on the foundation of binocular stereo vision theory, camera calibration, feature point extraction, image processing,3D matching and3D reconstruction, and other aspects of the technology having been further researched and discussed, and it has designed and implemented a set of rules for box shape of the object non-contact measurement system. The first stage of this article is image-forming principle of camera, introducing the basic imaging model and process of camera. Discussing various camera calibration algorithms on that basis, the author calculates internal and external parameters of camera through using Zhang Zhengyou calibration algorithm. And then, this article introduces epipolar geometry and a line constraint, and gives some related concepts of extreme line calibration and various correction algorithms. Through a further study of the conception and function of extreme line calibration, this article creatively proposes to apply extreme line calibration theory to calibration process of camera, that is to say, using the results of extreme line calibration to revising the used for the camera calibration image group, and then calibrating the camera a second time with the corrected image group. In the next stage, this paper discusses the research of image processing algorithm, feature point extraction algorithm and matching algorithms, and it realize target angular point extracting module, image preprocessing module and feature point extraction and matching module based on the functions which the system needs. In the last stage, this article goes into many kinds of3D reconstruction, basing on the existing algorithm, combining with the respective characteristics of triangle method and common perpendicular method, and it puts up with a3D reconstruction algorithm method (two-dimensional plus one-dimensional), using this method to calculate the position information of each vertex of this object. This system is designed and implemented under the VC++6.0programming environment, through a large number of experimental data, and the system shows that it can basically satisfy the meets of precision and true time measure of object volume. |
