摘要
以开发酶联斑点自动检验分析仪为目标,研究酶联斑点显微图像的自动识别计数技术。针对源图像背景复杂、边界模糊的特点,采用自动阈值边缘提取算法对源图像进行分割和轮廓提取,并基于酶联标板单元孔的几何特征筛选边缘轮廓,然后拟合图像的实际边缘轮廓,实现对目标区域的提取。通过改进的基于距离变换的分水岭算法对目标区域中的斑点进行分割,对分割后的斑点进行连通域划分并识别计数。实验结果表明:酶联斑点自动计数高效精准,结果判读综合准确率为95.31%,可替代人工镜检。
With the goal of developing an enzyme-linked spot automatic inspection analyzer,the technology of automatic recognition and counting of enzyme-linked spot microscopic images is studied.In view of the complex background and blurred borders of the source image,an automatic threshold edge extraction algorithm is used to segment and extract the source image,and the edge contour is filtered based on the geometric characteristics of the cell holes of the enzyme linked standard plate,and then the actual edge contour of the image is fitted to realize the extraction of the target area.Through the improved watershed algorithm based on distance transform,the spots in the target region are segmented,and the connected spots are segmented and identified by the segmented spots.The experimental results show that the automatic counting of enzyme-linked spots is efficient and accurate,and the overall interpretation accuracy is 95.31%,which can replace artificial microscopy.
作者
熊国顺
张从鹏
谢佳成
解毅
张堉晨
XIONG Guoshun;ZHANG Congpeng;XIE Jiacheng;XIE Yi;ZHANG Yuchen(College of Mechanical and Material Engineering,North China University of Technology,Beijing 100144,China)
出处
《机械工程师》
2020年第11期47-48,51,共3页
Mechanical Engineer
关键词
图像处理
斑点计数
边缘提取
分水岭算法
image processing
spot counting
edge extraction
watershed algorithm
