Research Of Distance-Measurement System Based On Binocular Vision

Nowadays with the increase of international terrorist explosion activities, countries all over the world are paying more attention to studying on field of explosive-handling robot. How to improve success ratio of work is an very important subject for scientific workers in many countries. Fixing the distance of static explosive and explosive-handling robot precisely is key technology for the explosive-handling robot to grab the static explosive, in hence, study on the interrelated measurement technology has very important practical roles. To meet the special need of work environment for explosive-handling robot, a semiautomatic binocular vision system based on the binocular stereo vision technology to measure the distances of targetsThis system designs two cameras modeling people’s eyes to take photos, mainly uses the imaging mechanism of people’s eyes, then it measures the distances of targets combined with the binocular stereo vision theory. Finally, results could be transferred to the master console through the serial ports, so as to guide explosive-handling robot’s work. The main contents and achievements would be introduced as following:1. The research status of binocular vision technology both at home and abroad was studied, the basic theories of binocular vision technology was analysed, the mathematical model of binocular vision system was deduced in detail, the structure of the embedded system included five parts, i.e., image acquisition, image pretreatment, camera calibration, feature extraction and stereo matching. All these laid theoretical groundwork for the following system design.2. Three kind of reference coordinates in the camera calibration and relationships among them were described. The traditional pinhole model was selected as the reference model for camera calibration, and the calculation methods for the right-and-left cameras in the binocular vision system were analysed in detail. Feature extraction and feature matching are the most important in the binocular stereo vision system, SIFT and Harris were analysed for further study. RANSAC algorithm was used, the matching rate was improved via distance information. In real environments, magic square was corner detected and matching analysed, to compare the above two algorithms.3. Embedded systems show excellent developing trend. A embedded hardware platform based on S3C2440 was established. The hardware structure in the system contained processor, camera, storage management, power management, LCD touch-screen and UART interface. The embedded Linux operating system was chosen to manage the software and hardware resources of the system. Qt/Embedded graphical user interface was taken to realize the alternation design between man and computer. Embedded Linux was ported to the platform according to binocular vision system, and application program were developed under Linux, mainly including image acquisition design, serial communication design and binocular ranging algorithm.Finally the precision of the binocular vision system was verified through experiments. The results indicate that the measured results reach the expected acquirements basically. The system possesses stable operation and accurate results. All these lay solid foundations for the subsequent development of vision system of explosive-handling robot.