图像聚焦判别函数的选择是自动图像测量中获取高质量图像的依据, 对几种用于自动图像聚焦判别的数字图像处理算法进行了较全面的性能比较, 定量分析了不同算法的计算速度、唯一性、准确性和灵敏度等。分析结果表明:梯度向量平方函数、拉普拉斯算法和二级梯度平方算法在单值性和灵敏度方面都较好; 罗伯特(Robert)梯度算法和梯度向量模方算法函数的稳定性比较好。以上的图像聚焦判别函数特性分析结果对于图像自动测量所需要的自动调焦控制具有指导意义。

The choice of the imaging focusing criterion function is the key of high-quality image acquiring in auto image measurement. The performances of several digital image processing methods used for autofocusing criterion are given. The focusing evaluation functions are compared and analyzed quantitatively through experiments in the aspects of calculation speed, unique value, precision and sensitivity. The results show that gradient square algorithm, Laplacian algorithm and second order gradient square algorithm have good behavior at the unique value and the sensitivity; and that both Robert algorithm and gradient mode algorithm have good behavior at the stability and calculation speed. These analysis results of image focusing criterion functions are useful for autofocusing in the automatic image measurement application.