Contour following parallel thinning for simple binary images

Thinning is a process of reducing the width of an image to just a single pixel. The resulting thinned pattern is useful for topological analysis and classification for the shape of patterns containing elongated or line-like parts such as printed or handwritten characters. Some existing well-known thinning algorithms were studied. Based on the information obtained from the study, we designed and implemented a new thinning algorithm known as CFPTA. The algorithm uses contour following parallel processing and a lookup table for the determination of pixel deletion. The processing time of CFPTA is fast in general. Although the skeletons generated are not perfect, edges are shortened, and all other characters of the images are preserved. This is adequate for applications of simple binary images such as character recognition.