二值图像连通域标记优化算法定义.pdfVIP

Optimization on algorithm of labeling connected components in binary images 1 1 2 LUO Zhi-Zao ZHUO Ying-Wu LUO Zhi-Shi (1 ．department of electronic engineering in Minjiang University in Fu Zhou City 350108 ； 2．Sanming City Electric Power Bureau 365000) Abstract: The improved algorithm is introduced based on analysis of existing algorithms of labeling connected regions through scanning pixels in binary images.The algorithm has many advantages such as faster,simpler,easier to realize than the others, has more preponderance in terms of multi-objects labeling,and keeping the original edges.The improved algorithm could label background as objects, label and merge connected regions via the two procedure of scanning binary image and scanning buffer block of unmerged regeions.It adopts sequential storage structure instead of chain,stack and tree structure,runs more rapidly, occupies much less memory than the unimproved algorithm. Key words :binary images;connected components;scanning by pixels;labeling Binary image is matrix which is component of white pixels and black pixels (commonly 1‘’ represents white pixels in binary image, as well 255‘ ’ in gray image)[1]. In binary images, labeling connected components indicates to label the connected pixels as one object according with the same connected rule(such as 4-Neighbour,8-Neighbour,m-Neighbour)[2].It is important to construct appropriate data structure,for the object ’s attribute and temporary operating result,such as wich object current pix