Abstract:Separating objects touching and overlapping with each other into single ones is very important for the counting and feature extracting.Most of the existing sepa rating algorithms are based on two hypotheses:(1)the touching objects tend to fo rm an acute angle on the contour where objects touched,and/or (2)there would al w ays be a lower optical density at touching points.For colony image,the bacteria cells vary in sizes and touching manners.Many touching cells are short of above hypotheses.A new separating algorithm based on 8-connected contour tracking is proposed in the paper.The boundary of touching-cell was tracked and deleted lay er by layer,and judgment was made by the tracking result to determine whether t h e touching-cell should be separated.The experiment results show the success rat io to separate bacteria touching-cell is 10% higher than the existing algorithm s.

A New Algorithm for Separating Cell in Bacteria Image