中文摘要：

提出了一种基于地面图像和卫星图像集成的火星车定位新方法，该方法利用由导航相机立体图像得到的火星表面的点云数据提取石块，同时对高分辨率卫星图像局部灰度统计提取石块，通过地面和卫星图像中石块分布模式的匹配，实现火星车在卫星图像中的定位，从而消除仅利用地面传感器和影像定位产生的累计误差。采用NASA勇气号火星车在多个摄站获取的地面图像以及HiRISE卫星图像进行了实验验证，结果表明这种方法在石块较多的地区能够取得很好的自动定位结果，定位误差小于HiRISE卫星图像的一个像素（0．25m）。

英文摘要：

This paper presents a new approach to rover localization based on integration of ground and orbital images. In this approach, rocks are extracted from 3-D point cloud data derived from the Navcam stereo images and are also extracted from high resolution orbital image through local image statistics; through matching the rock distribution patterns between the two kinds of images matching, the rover is localized in the orbital image, thus eliminating location error accumulation caused by only using ground sensors and images. The effectiveness of the proposed approach is demonstrated through experiments using Navcam images from multiple sites of NASA＇s Spirit rover and a HiRISE orbital image. The results show that it generates satisfactory results in rockabundant regions with a precision of higher than one HiRISE pixel （0.25 m）.