一种基于多尺度边缘提取的陨石坑检测算法

A Craters Detection Algorithm Based on Multiscale Edge Extraction

针对星体表面的陨石坑可用于探测器的自主导航、障碍识别等任务,提出一种基于多尺度边缘提取的陨石坑检测算法。该算法首先利用高斯金字塔得到不同尺度的陨石坑图像;其次,针对不同尺度的陨石坑图像,利用EDPF边缘提取算法对陨石坑进行边缘提取,并连接关键边缘像素点为直线段来近似表示图像边缘;然后将具有相同偏转方向的边缘直线段连接成圆弧,并将有相似半径和中心的圆弧拟合成候选圆和椭圆;最后对候选圆、椭圆进行验证。该算法的优点在于,能够准确地检测出陨石坑,有较高的检测率,且对存在较多陨石坑的图像有较好的检测结果。

As for the craters on the surface of planets can be used to autonomous navigation and obstacle recognition,thecraters detection algorithm based on multiscale edge extraction is proposed. In this algorithm,different scale craters are detectedthrough the Gaussian pyramid. For the different scales of crater images,firstly,using edge drawing parameter free(EDPF)to getedge segment of crater,and connecting critical edge pixels into line segment for representing the edge segment;then joining theline segments with the same turn direction into circular arcs,and fitting circular arcs with similar radius and center into candidatecircles or ellipse,finally validating candidate circles and ellipses through calculating the number of false alarm(NFA).Especially,the algorithm can be used to detect craters accurately with high detection rate,and has good performances in imageswith many craters.