关键词: 减法聚类, 直线检测, 霍夫变换, 网格化

Abstract: To solve the problem of undetected-line and pseudo-line resulting from unsuitable manual voting parameter in gridding Hough transform, a non-voting-parameter gridding Hough ttansform based on subtractive clustering is proposed. Firstly a two-stage scan in monotonous way is adopted to make every linelet as long as possible, and then the subset of voting linelets is automatically determined by characteristics of the lack of long linelets and the abundance of short linelets. Finally a fault-tolerant voting process is realized by using subtractive clustering. Our experimental results show that the proposed algorithm has a fast execution speed, and without manual voting parameter, and is very good to avoid the problem of undetected-line and pseudo-line resulting from unsuitable manual voting parameter by combining it with subtractive clustering.

Key words: subtractive clustering, straight-line detection, Hough transform, gridding