火星探测器着陆过程中图像匹配的性能研究

摘要

This paper uses the estimated location parameters of the image matching in the landing process of the Mars probe and analyzes the effects of the real-time image rotation captured by the probe and the scale changes on the image-matching algorithm in order to find out a higher matching precision through applying appropriate rotation angles and scales.It takes the SURF algorithm as the research object,and combines it with the RANSAC algorithm to eliminate the false matching points and get the accurate matching points.It then gives the correct concept of accuracy for assessing the matching results more easily.The experimental results show the dynamic rules in terms of the real-time image rotation angles,the scale changes and the matching accuracy.The application of this research to the landing process of the Mars probe proves that the selection of the appropriate rotation angle and the scale of the real-time image can somewhat improve the image-matching accuracy.%以火星探测器着陆过程中通过图像匹配估计位置参数为背景,针对探测器拍摄的实时图旋转以及尺度变化等对图像匹配算法的影响做一系列研究,寻找适当的旋转角度以及尺度大小来获得较高的匹配精度.以SURF算法为研究对象,结合RANSAC算法对误匹配点进行剔除,提纯匹配点,综合定义了正确率的概念,方便对匹配结果进行判断.实验结果表明,实时图旋转角度以及尺度大小和匹配的正确率呈一定的规律在变化,在应用到火星探测器着陆过程中实时图的选取时,相对于参考图选取适当旋转角度和尺度的实时图能在一定程度上提高图像匹配的准确率.