体视显微镜视频的快速校正算法

立体校正是双目视觉系统中必不可少的环节,这一环节包括了对特征点的提取和图像的校正。提出了一种针对十字标尺的校正方法,对于给定一幅图像中的一点,寻找另一幅图像中的对应点,这两点为空间同一场景坐标的投影;考虑了由于校正板直线不光滑而导致的特征点错误提取的问题,首先找到十字线上四个点,再间接找到实验需要的特征点。实验结果表明,该算法需要的参数数量少,实现速度快,最终实现的误匹配率低。适用于利用十字标尺求取参数的立体校正。

Fast correction algorithm of stereo microscope video

Binocular calibration is an essential part of the stereo vision system, it consists a feature point extraction and image correction. A method for cross-scale correction is presented. For a given point in an image, the corresponding point in another image is found, and the two points are the projection of the same scenarios. Considering that the straight line in the calibration board is not smooth, which will cause the mistaken extraction of feature points, four points on the cross line are first located in order to indirectly find the feature points needed in the experiment. Experimental results show that the number of parameters required in the algorithm is small, the speed faster, and mismatch rate much lower. The method is applicable in the three-dimensional correction using cross-scale.