月球车视觉系统立体匹配算法

Stereo matching algorithm for lunar rover vision system

为了满足月球车视觉系统实时性和可靠性的要求,提出了一种基于点和边缘特征提取的立体匹配算法。首先采用高斯滤波和有限对比适应性直方均衡化(CLAHE)方法对校正后的立体图像对进行预处理,以削弱噪声和对比度的影响。其次研究了图像的SIFT特征点和边缘特征提取和匹配方法,以获得月球车周围的环境概貌和障碍物信息。最后将匹配成功的SIFT特征与边缘特征相结合,计算特征点对应的三维信息,实现月面环境地形重构,验证了算法的有效性。

To meet the requirement of the lunar rover vision system on real-timeness and robustness, a new stereo matching algorithm was proposed based on extraction of point and edge features. The rectified stereo images were preprocessed by Gauss filter and CLAHE method to weaken the effects of noise and contrast. The extraction and matching methods of SIFT key point and edge features were studied to get the environment general picture and obstacle information around the lunar rover. The 3- D coordinates corresponding to feature points were calculated combining the matched SIFT features and edge features to realize the 3-D reconstruction of lunar surface environment terrain. Experimental results validated the effectiveness of the proposed matching algorithm.