一种基于特征空间的月球撞击坑自动识别算法

撞击坑不仅是月球表面常见的一种地质单元，也是研究月表地质年龄的一个重要因素，所以准确的识别撞击坑就很重要．本文提出了一种基于特征空间的对撞击坑进行自动识别的算法．新算法先将CCD图像转换成特征空间图像，在特征空间中通过自适应的区域检测寻找撞击坑候选区域．再对候选的区域进行椭圆拟合以进一步确认撞击坑，最后统计得到撞击坑个数．新算法在嫦娥一号采集的月球CCD图像上进行测试，测试结果显示该算法对月球中小撞击坑有较高的识别能力．

An automatic algorithm for detecting lunar impact craters in a defined feature space

Lunar crater is not only the most common geological unit on the surface of the moon, but also an important factor to study lunar surface＇s geological age. So how to detect crater accurately is very important. This paper presents an automatic algorithm for detecting middle or small size lunar impact craters which has been proposed in a defined feature space. The algorithm converts the CCD images into feature space images first. Then it will try to find the possible crater areas in feature space images. After those possible crater areas have been found, the edges of potential craters will be fitted, and the crater will be found if the fitting result is satisfied. The new algorithm has been tested in Chang＇E-1 data; and testing results have shown that more middle and small size craters could be detected by the new algorithm proposed.