基于平行双目立体视觉的测距系统

Distance measurement system based on parallel binocular stereo vision

基于计算机视觉的理论基础,在PC机上搭建出了平行双目立体视觉测距系统的实验平台,在前人的基础上对传统的标定方法进行了改进,实现了双目摄像机的标定,并采用特征点匹配领域的热点与难点算法尺度不变特征变换(SIFT)特征匹配算法实现了立体匹配,恢复出物体的深度信息。通过实验与计算,最终验证了该方法理论分析的正确性及实践的可行性。

Based on the computer vision theory, the experimental platform for the parallel binocular stereo vision distance measurement system was build on PC. The traditional calibration method based on the predecessors was improved, and binocular cameras calibration was accomplished. The stereo matching was accomplished by using Scale-Invariant Feature Transform （SIFF） matching algorithm which is key point of the feature points matching, the depth of the object was recovered. Through experiments and computing, the correctness of theoretical analysis and practical feasibility of the method is validated.