一种强粘连细胞图像的分割方法与实现研究

Research on Segmentation Method and Implementation of Images of Strong Adhesion Cells

当多个细胞严重粘连时,传统的分割方法难以取得理想的分割效果。针对大部分细胞呈近似圆形的特点,提出了一种基于Hough变换圆检测的粘连细胞分离方法:采用彩色梯度法提取粘连目标边缘,然后针对细胞的圆形特征,以边缘有效像素点与对应圆周长的比值作为分割阈值做圆检测确定各细胞圆心及半径,并建立实体细胞判别准则作为粘连分割实现依据,最后根据各细胞有效弧含像素的多少确定对应圆的完整性实现粘连分割。实验表明,与采用传统的分水岭算法比较,该方法在强粘连细胞的分割方面,具有明显的优势。

When cells adhere seriously, the traditional segmentation methods are difficuh to achieve the ideal segmentation result. Aiming at the features that most of cells are approximately circular, a separation method for adhesion cells based on the Hough transform on circle detection is put forward. The information of the target edge is extracted with the help of color gradient method and then, based on the circle features of the cells, the ratio of the edge＇s effective pixels to the corresponding circumference is set as the division threshold of circle detection to determine the center and radius of the cells. The solid cell criteria are established to judge the segmentation achievement. Finally, the integrity of the corresponding circle is determined and the segmentation implemented according to the number of pixels of the cells effective are. The simulation indicates that, compared with the traditional watershed algorithm, this method has obvious superiority in the segmentation of strong adhesion cells.