摘要
主要对单个宫颈细胞中的细胞核与细胞质边缘进行精确提取。为了让GVF Snake模型提取出细胞图像边缘,首先需要得到初始轮廓。对于细胞质,通过Otsu法粗分割得到初始轮廓;对于细胞核,由于被分割物体与背景的相对大小差距过大时,Otsu法不能正确地进行分割,因此使用梯度边缘信息改进的全局阈值法粗分割得到细胞核初始轮廓。通过放置精确的初始轮廓,提高GVF Snake的效率和准确度。
The nucleus and cytoplasm margin of a single cervical cell are extracted precisely. In order to obtain the edge of cell image by using GVF Snake model,firstly it is required to get initial contour. For the cytoplasm,the initial contour is obtained by Otsu method. For the nucleus,the Otsu method can not be segmented correctly due to the fact that the relative size between the divided object and the background is too large. Therefore,the initial contour of the cell nucleus is extracted by using the improved global threshold method with gradient edge information. By placing an accurate initial contour,the efficiency and accuracy of GVF Snake can be improved.
作者
孟楚楚
赵立宏
MENG Chuchu;ZHAO Lihong(School of Electrical Engineering,University of South China,Hengyang Hunan421001,China)
出处
《智能计算机与应用》
2019年第2期28-32,共5页
Intelligent Computer and Applications
基金
南华大学研究生科学基金项目(2018KYY005)
