一种基于图像处理的双目视觉校准方法

摘要

双目视觉是利用机器视觉进行障碍物检测的研究热点。针对双目视频不同步，导致立体匹配不精准的问题，提出了一种基于图像处理的双目校准算法。算法首先根据道路的先验特征模型，建立视觉校准的敏感区域，以减小计算量。然后对图像的灰度进行聚类，并通过边缘检测和形态学处理来检测道路中的目标物，从而确定动目标与固定物的距离，搜索同步视频帧，实现双目校准。室外真实场景的实验结果表明，该算法可以较好地校准双目视觉，从而使立体匹配更精准。%Binocular vision is a hot area of obstacle detection. A calibration method of binocular vision based on image processing is proposed in this paper. It is used to solve the problem of the asynchronism of two videos , which results in the inaccuracy of stereo matching. Firstly, in order to reduce the amount of calculation, an area of interest is gotten based on the priori feature model of the road. Then targets in the road are detected through clustering image gray scale , edge detection and mathematical morphology processing. Finally, two videos are calibrated according to the distance of moving target with the fixed one. Experimental results in real outdoor scene show that this method can effectively calibrate the two videos and make stereo matching more accurate.