摘要
针对自动聚焦过程中的聚焦评价函数、聚焦窗口、搜索算法进行了研究。首先改进了Robert函数增加了聚焦评价函数的陡峭度;其次针对固定聚焦窗口无法准确找到聚焦物体的缺点,提出了一种动态选择聚焦窗口的方法,该方法将图像分块,利用不同子块的梯度变化程度区分物体和背景,该方法更具适应性;进而提出了大步长和小步长相结合的爬山搜索算法,经过改进后的爬山搜索算法能更准确地找到焦平面;最后通过自行研发的显微视觉系统验证了所提自动聚焦方案的有效性。
For the focus evaluation function, focus window and focus searching algorithm in auto-focusing processing are researched. Firstly, the Robert function has been improved as the evaluation function to increase the steepness. Secondly, for the fixed focusing window cannot find focusing objects accurately, a new method of dynamic selection focusing window is proposed, which divides the image into blocks, and uses the gradient variation of different sub-blocks to distinguish the objects and the background, the method is more adaptive. Then, the improved mountain climbing method which combines the large step and small step has been proposed, and the micro-vision system can find the focal plane more accurately by this method. Finally, experiments demonstrate the effectiveness of the method.
出处
《计算机工程与应用》
CSCD
北大核心
2016年第14期197-201,共5页
Computer Engineering and Applications
基金
河北省自然科学基金(No.F2012203111)
河北省高等学校自然科学研究青年基金项目(No.2011139)
中国环境管理干部学院科研基金(No.BJ201604)
关键词
显微视觉系统
自动聚焦
评价函数
动态聚焦窗口
爬山算法
micro-vision system
auto-focusing
evaluation function
auto-focusing window
climbing algorithm
