A heuristic method for initial dominant point detection for polygonal approximations

Abstract

This paper presents a novel initial dominant point detection technique for polygonal approximation in digital planar curves. This technique concentrates on the local and global deviation of the curve and detects the dominant point of the polygon. An iterative split and merge strategy is used effectively to insert and/or delete vertex during the approximation of the polygon. Since the internal steps are automated, a symmetric and better approximation is achieved. The technique is robust to rotation and noise of the shape and produces better results compared to the results obtained by recent work. The performance of the proposed system is evaluated using the benchmark data set and the same is compared in terms of the quantitative and qualitative measures. The experimental results show that proposed technique has outperformed an existing similar method with respect to visual perception and numeric metrics.