基于互相关的自动聚焦方法

Auto-focusing algorithm based on cross correlation

针对自动聚焦系统,提出了基于互相关的自动聚焦算法.分析了传统自动聚焦算法在速度和精度方面存在的不足,引入互相关原理,定义基于互相关的模板相关算法FTC.由于选择模板图像的不同,实验的结果不同,进而导致自动聚焦分析的算法也不相同.根据模板图像和实验结果,得出该算法存在2种形式———模板相关的第1种形式和第2种形式.模板图像为聚焦前位置图像,称为模板相关的第1种形式,该形式的实验曲线是由小到大,再减小的过程;模板图像为聚焦图像时,称为模板相关的第2种形式,该形式的实验曲线是由大到最小,再增大的过程.2种形式的图像处理实验结果,与其他几种主要聚焦评价函数(平方梯度函数,Brenner函数)进行比较,结果表明该算法简单,原理容易实现,实验曲线能避免各种波动,因此基于互相关的自动聚焦算法提高了自动聚焦的精度和速度.该算法应用于微流控芯片对准装配自动聚焦系统中,取得了良好的聚焦效果.

Aiming auto-focusing system, auto-focus algorithm based on cross correlation was brought forward. The limitation of conventional auto-focusing in definition and velocity was analyzed. Correlation principle was introduced, and template correlation algorithm based on cross correlation was defined. Owing to difference of template image selected, the result of experimentation is different, so that auto-focus analysis algorithm is different. Based on different template image and the result of experimentation, two forms of correlation function were presented, including the first template correlation and the second template correlation. If the place of template image is before focusing, the form is named the first form, curve of which changes from small value to big, then to small, if the template image is focusing image, the form is named the second form, curve of which changes from big value to least, then to big. The result which two forms processes image was compared with the result of experimentation of other focus evaluating functions, such as square gradient function, correlation function. By experimentation comparison, it concludes that the auto-focusing algorithm based on cross correlation improves definition and velocity of auto-focus algorithm. The algorithm has been successfully tested to be effective on micro-fluidic alignment assembly auto-focusing system. It shows that the algorithm has achieved better auto-focus effects.