An algorithm of Euclidean distance transform was proposed by raster scan in obstacle space. It propagates distances line by line

checking the visibility of target pixels

and calculating the distance by finding a shortest way from a source to the target keeping away from obstacles. The algorithm is straightforward for distance calculation

involving in no complicated data structure

and having a linear computational complexity. Case study showed that the calculated distances by this algorithm were accurate

better than the results by the distance accumulationbased algorithm

and suitable for sources and obstacles in any types of shapes

including point

line and area objects.