细胞工厂光学显微成像与图像处理技术

针对细胞工厂监控系统长工作距和大倾斜角的观测需求,设计了一种结构简单、成像清晰的光学显微成像系统.由于获得的细胞图像光照和色彩不均、样本浑浊、细胞重叠、边界粘连较多、细胞间距不明显,采用单尺度Retinex算法进行图像预处理,并结合Otsu阈值分割法与K均值聚类法进行细胞图像分割处理,最后应用改进的快速连通区域标记以及高准确度细胞计数方法进行细胞个数统计.仿真实验和实际测试结果表明:该系统成像分辨率和清晰度均达到工程需求,能够较清晰地辨识出培养皿中细胞的形态和分布情况.细胞显微图像处理方法取得了良好的图像增强效果,弱化了由光照不均和样本浑浊造成的人眼视觉无法清晰分辨细胞的现象,消除了由于图像分割不到位造成统计误差,细胞计数准确度达到95%以上.该方法适合多种类型细胞监测与计数处理,可满足细胞工厂实时准确监控的要求.

Accuracy Analysis of Gravity Model Based on Fictitious Compress Recovery Approach

The potential field determined based on the fictitious compress recovery approach is influenced by the errors contained in the boundary （the Earth＇s surface or the surface corresponding to the satellite altitude） values. Given the boundary value with definite accuracy, the accuracy of the field determined based on the fictitious compress recovery approach is estimated, and it is theoretically shown that the determined field has the same accuracy level as the given boundary value.