简单图像的快速聚焦  

Fast focus on simple images

作　　者：陈芳[1] 张存继 韩延祥[1] 史金飞[2] 

机构地区：[1]东南大学机械工程学院,江苏南京211189 [2]南京工程学院,江苏南京211167

出　　处：《光学精密工程》2014年第1期220-227,共8页Optics and Precision Engineering

基　　金：国家自然科学基金资助项目(No.51275090);江苏省科技成果转化专项资金资助项目(No.BA2010093);江苏省科技支撑计划资助项目(No.BE2008081)

摘　　要：为了有效地实现简单图像的快速高精度自动聚焦,提出了一种新的快速聚焦算法。首先在相机进行调焦时自动获取系列零件图像,并记录各图像对应的位置;然后计算每张图像中目标边缘的灰度变化跨度值,从中找到跨度值最小的图像,即为最清晰图像;最后把其所在位置反馈给硬件驱动系统,实现自动聚焦。用新算法分别对不同形状和不同材料的零件,在添加椒盐噪声和没有添加噪声的情况下进行实验,并与计算量小的几种经典最优聚焦函数做了对比试验。结果表明,用新算法对简单图像进行聚焦比常用的最优聚焦函数更敏锐,单峰性更好,抗噪能力更强,而且速度比最快的绝对梯度函数快30%以上。因此,新算法在拍摄简单场景时,鲁棒性好,可以更好地实现快速聚焦。To quickly realize the autofocus of a simple image and further improve the efficiency of automated part fabrication, a novel fast focusing algorithm was proposed according to the characters of images with single and clear backgrounds and foregrounds. Firstly, a series of part images including their position information were acquired. Then, all variance grey span values of target edges in each image were calculated automatically and the minimum span value of the clearest image was found quickly. Finally, the position information of the clearest image was feedbacked to a hardware driving system to complete the autofocus. An experiment on several parts with different shapes or materials was performed and obtained results were compared with that of the classic optimal focus function. The experimental results on those images added with salt-and-pepper noises demonstrate that the new algo- rithm is not only more sensitive but also has stronger unimodality and stronger anti-noise ability than those of classic optimum algorithms. Moreover,its computing speed is more than 30% faster than that of the fastest absolute gradient function. It concludes that the novel algorithm is robust and can be better used for a fast focus on simple scenes.

关 键 词：自动聚焦 目标边缘 跨度值 聚焦函数 鲁棒性 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]