Abstract: Moon environment exploration is very important to moon vehicle autonomous navigation on moon surface. Stone as a kind feature of moon surface, it will be a kind of obstacle for moon vehicle to run. The paper obtains gradient image by the morphological gradient, and it is segmented by the minimum error method. Binary image is filtered by utilizing morphological filter. Then, using the information of stone shadow and projection transform, detection and position of stone is utilized in part environment.

Key words: Moon environment, Stone detection, Morphological gradient, Minimum error

摘要： 月球环境感知是未来月球车在月球表面自主导航的关键。石块作为月球表面的一种地理特征，将成为月球车行驶的障碍物。该文从获取梯度图像研究出发，提出了基于最小误差分割原理的石块图像分割方法。在获取分割后的二值化图像中，采用了形态学滤波算子进行去噪，然后利用二值图像投影变换，并参考石块底部阴影位置信息，辅助完成对部分场景下的石块识别和定位。

关键词: 月球环境, 石块识别, 形态学梯度, 最小误差