基于轮廓的血细胞图像拼接算法的设计与实现

作者：王晨，庞全 日期：2009-12-30/span> 浏览：3713 查看PDF文档

基于轮廓的血细胞图像拼接算法的设计与实现

王晨，庞全
(杭州电子科技大学 生物医学工程与仪器研究所，浙江 杭州 310018)

摘要：针对显微镜下血细胞切片图视野较小的问题，提出了一种基于轮廓的拼接算法。首先通过改进的迭代法分割出白细胞胞核；采用边界跟踪法提取胞核轮廓后，计算各轮廓图像的不变矩，并构建不变矩的欧式距离矩阵；根据最近邻原则找出匹配轮廓，以匹配轮廓的重心为控制点计算仿射坐标变换系数。实验结果表明，该方法具有旋转、平移、缩放（RTS）不变性，且精度高，自动化程度好。
关键词：图像拼接；迭代阈值法；轮廓；不变矩；最近邻原则
中图分类号：TP391.41文献标识码：A文章编号：1001-4551(2009)11-0044-05

Design and realization of blood cells image mosaic algorithm based on contour
WANG Chen, PANG Quan
(Institute of Biomedical Engineering and Instrument, Hangzhou Dianzi University, Hangzhou 310018, China)
Abstract: Aiming at the problem of small visual field of blood cells under microscope, a new mosaic algorithm based on contour was proposed. First, segment blood cell nucleus by improved iteration threshold segmentation approach, after contour extraction with boundary tracking, calculate invariantmoment of every contour image and construct invariantmoment euclidean distance matrix. According to nearest neighbor rule, find matched contours, and then calculate affine transformation parameters with their corresponding centers of gravity. The results of experiments show that the method could be invariant to translation, rotation and scaling(RTS), and has high precision and automation.
Key words: image mosaic; iteration threshold approach; contour; invariant|moment; nearest neighbor rule