Size-invariant four-scan Euclidean distance transformation

摘要: Abstract Distance transform (DT)(1) is used to convert a binary image that consists of object (foreground) and nonobject (background) pixels into another in which each pixel has value corresponding the minimum distance from background by predefined function. The Euclidean more accurate than others, such as city-block, chessboard chamfer, but it takes computational time due its nonlinearity. By using relative X Y coordinates computed source mapping neighbors well correction particular cases, transformation (EDT) can be correctly obtained just four scans an image. In other words, new algorithm achieves complexity EDT linear size