In-Place Linear-time Algorithms for Euclidean Distance Transform

Given a binary image, Euclidean distance transform is to compute for each pixel the Euclidean distance to the closest black pixel. This paper presents a linear-time algorithm for Euclidean distance transform without using any extra array. This is an improvement over a known algorithm which uses additional arrays as work storage. An idea to reduce the work space is to utilize the output array as work storage. Implementation results and comparisons with existing algorithms are also included