查询结果:   叶含笑,傅斌,夏炳江.HE染色病理切片最优二值化算法研究[J].计算机应用与软件,2014,31(5):215 - 218,222.
中文标题
HE染色病理切片最优二值化算法研究
发表栏目
图像处理与应用
摘要点击数
664
英文标题
ON OPTIMAL BINARISATION ALGORITHM OF HE DYEING PATHOLOGICAL SECTION
作 者
叶含笑 傅斌 夏炳江 Ye Hanxiao Fu Bin Xia Bingjiang
作者单位
浙江中医药大学 浙江 杭州 310053     
英文单位
Zhejiang Chinese Medical University,Hangzhou 310053, Zhejiang, China     
关键词
阈值图像分割 颜色欧氏距离 最大类间方差法 颜色深度迭代法
Keywords
Threshold image segmentation Colour Euclidean distance Maximum between-cluster variance Colour depth iteration
基金项目
浙江省教育厅2012年度科研计划项目(Y201224333);浙江省钱江人才计划项目(QJD0802007)
作者资料
叶含笑,副教授,主研领域:医学图像处理。傅斌,助理研究员。夏炳江,硕士生。 。
文章摘要
基于阈值的图像区域分割难点在于阈值的选择。目的: 为HE染色病理切片区域分割提供通用稳定的算法,对感兴趣区域进行二值化。方法: 选择前景色和背景色作为种子点,求取种子点的RGB值欧氏距离作为阈值,对图像二值化分割,与颜色深度的最大类间方差法、颜色深度迭代法等多种自适应获取的阈值二值化图像分割效果进行比较。结果: 基于背景色和前景色种子点选择的颜色欧氏距离对图像二值化分割稳定性和准确性最好。结论: 在区域划分合理的情况下,以种子点颜色深度的欧氏距离阈值进行二值化,可以提取到感兴趣区域。
Abstract
Threshold-based segmentation of image region has the difficult in threshold selection. Objective: Providing a general and stabilised algorithm for the segmentation of HE dyeing pathological section regions, and binarising the interested regions. Method: choosing the foreground and background colours as the seed points and calculating the RGB value Euclidean distance between the seed points as the threshold, then dividing the image in binarisation way; comparing it with the segmentation effects of several threshold binarised images adaptively obtained from the maximum between-cluster variance method of colour depth and the colour depth iterative method, etc. Results: the colour Euclidean distance based on background and foreground colours seed point selection has best stability and accuracy on image binarisation segmentation. Conclusion: under the circumstance of reasonable regional division, and binarising with Euclidean distance threshold of seed points’ colour depth, the interested regions can be extracted.
下载PDF全文