一种新的完全欧氏距离变换算法

A New Algorithm for Complete Euclidean Distance Transform

论文提出了一种基于边界剥离的二维完全欧氏距离变换算法。该算法从物体目标的最外层边界开始,自外向内、逐层对物体目标区域进行边界跟踪、剥离。在跟踪过程中,根据当前边界像素点的已获得距离变换结果或为背景的邻域像素信息,计算其与最近背景像素间的欧氏距离,从而实现距离变换。和已有算法相比,文中算法具有简单快速、容易实现,得到的是完全欧氏距离的优点,在分离粘连物体的应用中,取得了良好分离效果。

A 2-D complete Euclidean distance transform algorithm based on contour-stripped is presented in the paper.The algorithm tracks the contour of the object.During the tracking,the distance between the current object pixel and its nearest background pixel is computed on the distance transform result of its neighbor pixels obtained by previous tracking.After every tracking is done,the pixels of this lay of contour are deleted from the object and next tracking starts.This course is done repeatedly until the area of the object is empty.And at last,the distance transform of the whole binary image is obtained.Comparing with existing algorithms,it is simple,computationally inexpensive and easy to be implemented;moreover,it can get the complete Euclidean distance.As a result,its application in the separation of touching object is perfect.