【作者】 李阳超； 熊显名；

【Author】 LI Yang-chao, XIONG Xian-ming, YI Xi-peng Guilin University of Electronic Technology, Eletronic Engineering Department, Guilin, Guangxi 541004

【机构】 桂林电子科技大学电子工程系；

【摘要】 在用计算机作数字图像采集和数字图像处理的自动聚焦方法中,由于图像采样、量化及传输过程中各种干扰引起的图像噪音,导致评价函数曲线可能存在多个极值。目前光学自动聚焦系统普遍采用盲人爬山比较法,当评价函数曲线存在多个极值时,采用这种方法可能会将一非最大值的极值判定为聚焦点。针对这种虚假聚焦点的问题,本文提出了一种将每帧图像的聚焦函数值存于一表中,然后通过查表找出聚焦函数值的最大值的方法。在选取Tenengrad函数作出评价函数的基础上,采用查表法在PC 机VC平台下,建立基于数字图像处理程序控制的自动聚焦系统,经过大量的实验分析,证明这种方法能很好的解决盲人爬山比较法中存在的虚假聚焦点的问题且聚焦精度较高。更多还原

【Abstract】 In auto-focusing system, computer is used to gather and process digital image. Several extrema may be produced in evaluating function curves, as image noise is caused by interferences in image gathering, quantifying and transmission processes. Nowadays, blind-man climbing compare method is generally used in optical auto-focusing system. But when several extrema exist in evaluating function curves, a false extremum, not maximum, is selected as focus point. In the paper, we propose a method to resolve the problem: focus function values of each image are stored to a table, and the maximum is found by looking up the table. With this method and VC studio on personal computer, we build auto-focusing system controlled by digital image process program, while Tenengrad function is selected as evaluating function. Many experiment analyses are done to testify that the method is feasible to resolve the problem of false focus point, and system’s focusing precision is high.更多还原