Earticle

Skeletonization has been a part of morphological image processing for a wide variety of applications. Skeletonization algorithms have played an important role in the preprocessing phase of OCR systems. Many algorithms for vectorization by skeletonization have been devised and applied to a great variety of pictures and drawings for data compression, pattern recognition and raster-to-vector conversion. The vectorization algorithms often used in pattern recognition tasks also require one-pixel-wide lines as input. But parallel skeletonization algorithms which generate one-pixel-wide skeletons can have difficulty in preserving the connectivity of an image or generate spurious branches. In this paper an alternative parallel skeletonization algorithm has been developed and implemented. This algorithm is better than already existing algorithms in terms of connectivity and spurious branches. A few most common skeletonization algorithms have been implemented and evaluated on the basis of performance parameters and compared with newly developed algorithm.