针对光学透镜参数测量系统中的对焦问题，设计了一种粗精结合的复合自动对焦方案。在粗对焦过程中，选用全局图像的方差函数作为清晰度评价函数，用均值比较的爬山法实现焦面搜索，并用三点法判断初始对焦方向。在精对焦过程中，利用基于数学形态学的条件膨胀算法和形状因子提取目标图像，并以目标图像区域内的Brenner函数作为清晰度评价函数，用单点比较的爬山法实现焦面精细搜索。实验表明，该对焦方案不仅对焦精度高，还有较大的对焦范围，能够保证对焦位置包含在镜头景深范围内，且透镜参数的测量结果准确，稳定。

A composite autofocusing technique consisting of coarse autofocusing and fine autofocusing is proposed for test system of optical-lens parameter. In the coarse autofocusing procedure, variance function achieved from the whole image is selected as sharpness evaluation function, mean comparison method is applied to realize the hill-climbing algorithm for focal plane searching, and a 3-point method is used to initialize the searching direction. In the fine autofocusing procedure, a conditional dilation method based on mathematical morphology and a shape factor are employed to extract the connected regions, namely, the target images. Brenner function within the target image is selected as the sharpness evaluation function, and the hill-climbing algorithm with single-point comparison is used to find the focal plane accurately. The experiments validate that the composite autofocusing technique possesses not only high accuracy but also large autofocusing range. It can guarantee that the focusing position locates in the depth of field of the camera lens and the lens parameter measuring results possess good accuracy and repeatability.