Parallel implementation of a fast thinking algorithm using image compression

摘要

The multiprocessor implementation of a new thinning algorithm based on image compression is presented. In this algorithm, a binary image is compressed and then thresholded prior to applying the thinning process thereby reducing the width of edges. This results in fewer iterations of the thinning process being required to achieve a skeleton image. Rigorous connectivity checks using a set of 38 different masks are efficiently implemented to preserve the connectivity of pixels and shape of image features.