摘要
为了准确地确定聚焦平面位置,对自动聚焦技术中聚焦区域选择、聚焦评价函数的确定和搜索策略等关键部分进行了研究.针对聚焦区域,以一阶矩值为聚焦窗口中心,由具体处理的图像分辨率与图像内容决定聚焦区域范围;针对简单图片序列与复杂图片序列,对各种聚焦评价函数进行了仿真与验证;提出了一种改进的爬山法,充分利用聚焦评价函数指标陡峭区宽度与局部极值因子,同时设置两个搜索斜率阈值,为自适应修正搜索步长奠定了较好的基础.结果表明:Laplace聚焦评价函数与SML聚焦评价函数各方面性能优于其他聚焦函数;本文提出的搜索方法可以准确地搜索到焦平面的位置,可为后续基于图像分析操作的工业化与自动化奠定良好的基础.
In order to obtain accurate focus, three key contents for automatic focus are researched in this system, which include the choice of focus area, focus evaluation function and focus searching algorithm. Aiming at focus area, the center is determined by first order moment of the processed images. The size of focus area is based on the resolution and content of the processed images. Various focus evaluation function is simulated and verified based on serial simple images and complex images. An improved hill-climbing method is put forward, which takes advantage of the indexes of focus evaluation function, such as the steep width and local extrema factor. Two slope thresholds are set so as to adaptively modify step size. The reults show that： Laplace focus function and SML focus function are superior to other functions; this method can search the accurate focus site. Accurate autofocus system provides better basis for the industrialization and automation based on image analysis.
出处
《天津工业大学学报》
CAS
北大核心
2011年第6期53-57,共5页
Journal of Tiangong University
基金
天津市科技计划项目(10ZCKFSF01400)
河北省自然科学基金资助项目(F2008000750)
关键词
自动聚焦系统
聚焦策略
聚焦评价函数
斜率阈值
陡峭区宽度
局部极值因子
automatic focus system
focus searching algorithm
focus evaluation function
slope threshold
steep width
local extrema factor
