A Fast and Complete Thinning Algorithm for Character Image

Lei GUAN, Zheng-lin LI, Zhi WANG

Abstract


In order to refine the character image and realize feature extraction, an improved thinning algorithm of character image is proposed based on the research and analysis of the existing image thinning algorithm. The algorithm constructs the thinning templates and restoring templates based on 14 neighborhoods. The algorithm not only improves the effectiveness of matched direction of the reservation template, and completely refine the image, but also retains the advantages of other image thinning algorithm such as central axis and topology preserving. Experiments show that the new algorithm has a fast speed in the process of thinning, and owns the advantage in maintain 8 neighborhood connectivity. The thinning results of the new algorithm are smooth and without burr, which show that it is an efficient algorithm for fast and complete thinning.

Keywords


Thinning, Feature extraction, Template match, Skeleton


DOI
10.12783/dtetr/mcae2017/15960

Refbacks

  • There are currently no refbacks.