A New Scan-Line Algorithm Using Clustering Approach

Abstract

Correct recognition of the lines is essential for technical drawing understanding. Automation solution is quite difficult due to the limitations of machine vision algorithm. In order to promote development of better technology, according to the fast and high-quality clustering algorithm Particle Swarm Optimization (PSO), a new fast and high-quality line clustering algorithm present in this paper, that consisting of one scan-line connected components processing are clustered and an appropriate measure to recognize the pattern of every line including the dash-line in the drawing paper. The underlying mechanisms are excluding isolated components, a sequential stepwise recovery of components that meet certain continuity conditions and the results presented the node-tree structure that can enhance efficiency of computer. The performance of the algorithm is better in our experiment