基于方向梯度直方图的陨坑预检测算法

Pre-detection Algorithm of Meteor Crater Based on Histogram of Oriented Gradient

陨坑是行星软着陆导航与避障过程中的主要检测目标,目前陨坑检测方法需要被检测图像尽可能地包含完整且清晰的单一陨坑,因此对图像陨坑进行预检测处理得到了广泛关注。通过人工模拟地形方法建立行星陨坑图像数据库,提出一种基于灰度梯度直方图特征(HOG)的支持向量机(SVM)图像陨坑分类方法。该方法可以有效地提取图像中可能包含陨坑的图像子图,并作为后续精确标记陨坑位置的图像处理方法的输入数据。仿真结果表明,该方法能够有效地提高陨坑标记的准确性和实时性。

Meteor crater is the main target detected for planetary soft landing navigation and obstacle avoidance. The current craterdetection method requires the detected image to contain a complete and clear single crater. So the pre-detection processing of crater image has been widespread concern. In this paper,based on the meteor crater image database established by artificial terrain simulation method,a crater image classification method using support vector machine （ SVM） based on gray-scale histogram of oriented gradient （HOG） is proposed. This method can effectively extract the image subgraphs in the image that may contain craters,and can be used as input data for subsequent image processing methods that accurately mark the location of the craters.