Abstract
Thinning algorithms are widely used in many image processing tasks. Many thinning algorithms were proposed but they usually tend to process all image pixels in every iteration. Two approaches to contour thinning are described and a short discussion about their features is given. These approaches can be implemented as sequential or parallel algorithms with different deletion rules. Results of comparison and analysis are presented in this paper.
Highlights
Many image processing and pattern recognition problems use thinning as one of the processing step
Thinning algorithms are relatively fast compared to other skeletonization techniques, they are still slow for some tasks
As the number of contours decreases and the number of background and noncontour foreground pixels increases, CT1 algorithm became faster than Zhang-Suen thinning algorithm
Summary
Many image processing and pattern recognition problems use thinning as one of the processing step. These problems include vectorization of raster maps and engineering drawings [1], character recognition and analysis [2, 3] and more [4,5,6,7]. Like other segmentation techniques, thinning has its advantages and disadvantages. Thinning algorithms are relatively fast compared to other skeletonization techniques, they are still slow for some tasks. Many thinning algorithms were proposed [8, 9, 10, 13] but they usually tend to process all image pixels in every iteration
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
