Automatic Cell Image Segmentation Based on B-Snake Model with Constraint of Morphology

doi:10.11834/jig.20050107

Views : 0 下载量: 217 CSCD: 0

PDF
Export
Share
Collection
Album

Automatic Cell Image Segmentation Based on B-Snake Model with Constraint of Morphology
Vol. 10, Issue 1, Pages: 31(2005)
Published：2005
DOI： 10.11834/jig.20050107
稿件说明：

移动端阅览

DOI：

Automatic Cell Image Segmentation Based on B-Snake Model with Constraint of Morphology[J]. Journal of Image and Graphics, 2005, 10(1): 31. DOI： 10.11834/jig.20050107.

摘要

由于在细胞图像中经常出现细胞重叠的现象，从而给后续处理带来了困难，为了准确分离细胞，提出了一种自动分割方法，即首先使用一种可变腐蚀元的迭代腐蚀算法来产生改进的距离图，然后提出了受数学形态学约束的B样条活动轮廓模型，利用形态学方法初始化活动轮廓，最后通过该模型求出各细胞的准确边界。实验结果表明，该方法能有效地分离重叠细胞，并能准确定位细胞的完整边界。

Abstract

In cell images the clustering phenomenon frequently appears. In order to separate the clustering cells

an automatic segmentation algorithm based on a novelB2spline active contourmodel (B2Snakemodel) is proposed. The novel B Snake’s energy terms depend notonly on the cell image itself butalso on amodified nonlineardistance image. Firstly

the algorithm generates the distance image. The distance image is called a nonlinear distance image because of the nonlinear relationship between the gray levels ateach pixel and the distance from thatpixel to its nearest background pixel. To produce themodified nonlinear distance image

an iterative erosion method with a dynamic structure rather than a fixed one is used in the algorithm. Secondly

the B Ssnake model is initiated via morphological operation on the modified nonlinear distance image. Two initializationmethods

one is faster but the other ismore precise

can be chosen according to the cell types in the cell images. Finally

the novelB2Snakemodel obtains the cell boundaries under the effect of both the original image force and the modified nonlinear distance image force. Experimental results show that our algorithm is effective for automatic cell image segmentation.