This study integrates two different computer vision approaches, namely the circular Hough transform (CHT) and the determinant of Hessian (DoH), to detect automatically the largest number possible of craters of any size on the digital terrain model (DTM) generated by the Mars Global Surveyor mission. Specifically, application of the standard version of CHT to the DTM captured a great number of craters with diameter smaller than ~ 50 km only, failing to capture larger craters. On the other hand, DoH was successful in detecting craters that were undetected by CHT, but its performance was deterred by the irregularity of the topographic surface encompassed: strongly undulated and inclined (trended) topographies hindered crater detection. When run on a de-trended DTM (and keeping the topology unaltered) DoH scored higher. Current results, although not optimal, encourage combined use of CHT and DoH for routine crater detection undertakings.

这项研究集成了两种不同的计算机视觉方法，即圆形霍夫变换(CHT) 和Hessian 行列式(DoH)，以自动检测火星生成的数字地形模型 (DTM) 上任何大小的尽可能多的陨石坑全球测量员任务。具体来说，将CHT标准版本应用于DTM仅捕获了大量直径小于~50公里的陨石坑，未能捕获更大的陨石坑。另一方面，DoH 成功地检测到了 CHT 未检测到的陨石坑，但其性能因所包含的地形表面的不规则性而受到阻碍：强烈起伏和倾斜（趋势）的地形阻碍了陨石坑的检测。当在去趋势 DTM 上运行（并保持拓扑不变）时，DoH 得分更高。目前的结果虽然不是最佳的，但鼓励结合使用 CHT 和 DoH 进行常规弹坑探测工作。