【摘要】 月球表面环境感知是月球车自主导航的关键,月岩和月坑作为代表性的月面环境特征是月球车行驶过程中的主要障碍物。针对月球车自主导航和探测问题提出一种月岩和月坑的分割方法,首先对具有代表性的月面环境(月岩、月坑)进行各异向性扩散滤波(PM滤波),然后用最大类间方差法(Otsu)进行详细研究,对比分析最大类间方差法及基于最大类间方差法的两种改进方法。实验结果表明,基于粒子群的二维最大类间方差法在分割月岩和月坑图像时可取得较理想的效果。最后结合已有的三维重建平台,很好地在其基础上加入图像分割模块,为下一步避障和路径规划做准备。更多还原

【Abstract】 Lunar surface environmental perception is the key to lunar rover autonomous navigation. Moon rocks and moon pits are main obstacles as representative characteristics of lunar surface in the driving process of lunar rover. A segmentation method is proposed for lunar rover navigation and detection problems. The maximum interclass variance method( Otsu) is studied in detail after performing anisotropic diffusion filter( PMfilter). The maximum interclass variance method and two improved methods based on the maximum interclass variance method are compared and analyzed. Experimental results showthat Particle swarm algorithm based two-dimensional maximum interclass variance method can achieve better results for segmenting images of moon rocks and moon pits,which provides technical reserves for obstacles avoidance and path planning combined with the existing three-dimensional reconstruction platform.更多还原