一种显微镜自动聚焦算法  

A Focusing Algorithm in Automatic Microscope

作　　者：梁隆恺 赵晶[1] 何勇军[1] LIANG Long-kai;ZHAO Jing;HE Yong-jun(School of Computer Science and Technology,Harbin University of Science and Technology,Harbin 150080,China)

机构地区：[1]哈尔滨理工大学计算机科学与技术学院,黑龙江哈尔滨150080

出　　处：《哈尔滨理工大学学报》2018年第2期46-52,共7页Journal of Harbin University of Science and Technology

基　　金：国家自然科学基金(61673142;61471145;61305001)

摘　　要：传统的自动聚焦算法时间复杂度高,计算现在的高清图像速度慢,导致聚焦效率低下。针对这一问题提出了一种新的聚焦方法。该方法在分析灰度值与图像清晰度之间关系的基础上,提出了灰度非零值统计函数和低灰度值统计函数来评价聚焦清晰度,具有运算量小,评价准确的优点。然后采用两种评价函数结合的方式实现了一种变步爬山搜索策略。该策略在远焦点位置采用灰度零值比较评价函数和大步长快速粗略搜索焦点,在近焦点位置采用低灰度值统计函数和小补偿精细搜索焦点。实验表明,所提出的方法在聚焦效率和清晰度两个方面都优于目前的方法。The traditional autofocus algoritlim witli high time complexity has slow speed of current high-definition images that lead to low eficiency of tlie focusing.To solve this problem,we improve auto focusing by improving focusing evaluation and focus searching.First,two evaluation functions based on gray change,namely gray zero comparison method and low gray value statistics method,are put forwarded.Then the proposed evaluation functions are applied to realize the variable step climbing search strategy.The experimental results show that the proposed method in two aspects of focusing efficiency and clarity are better than current methods.

关 键 词：自动聚焦 自动显微镜 变步爬山 灰度非零值统计法 低灰度值统计法 

分 类 号：TH742[机械工程—光学工程]