“…The classical, binary DTs are useful, for example, for measurement and description of binary objects. Distance functions and transforms that are defined as minimal cost paths in general images include geodesic distance [9], fuzzy distances [19], and the minimax cost function; they have been used, for example, in segmentation and saliency detection [18]. The most common algorithms for computing distance transforms are: (i) raster scanning methods, where distance values are propagated by sequentially scanning the image in a pre-defined order [1,5]; (ii) wave-front propagation methods, where distance values are propagated from low distance value points (the object border) to the points with higher distance values, using data structure containing previously visited points and their corresponding distance values, until all points have been visited [15,21]; and (iii) separable algorithms, where one-dimensional subsets of the image are scanned separately, until all principal directions have been scanned [4,20].…”