关键词: 直线提取, Split-and-Merge, 特征描述, 激光传感器, 最小二乘法

Abstract: Aiming to the problem that the Split-and-Merge segmentation algorithm is very sensitive to changes on some parameters and inefficiency, this paper proposes a self-adaption line extraction algorithm based on it. This algorithm divides the laser sensor data into many near point aggregates by a self-adaption threshold, after that it segments the near point aggregates to many lines with Prototype-based fuzzy clustering algorithm. It estimates lines parameters according to the least square criterion. Experimental results demonstrate that all of robustness to line segmentation, precision to line extraction and efficiency to the algorithm are improved significantly.

Key words: line extraction, Split-and-Merge, feature description, laser sensor, least squares method