电子摄像式刀具预调测量仪的自动聚焦技术研究

Auto-focus Technology of Tool Presetting and Measuring Machine Based on Computer Vision

自动聚焦是电子摄像式刀具预调测量仪的重要模块,其中清晰度评价函数和聚焦点搜索策略直接影响到聚焦的速度和精度。通过对比分析典型清晰度评价函数的计算时间、灵敏度、单峰性等特性,得出Brenner函数和小波函数是刀具预调测量仪聚焦系统的最佳聚焦测度函数。为克服聚焦搜索运动中的机械传动误差,提高搜索的速度和精度,提出利用离散二阶差分预测模型来预测镜头的运动方向,大步长快速到达近焦区域,然后缩小步长,利用曲线拟合算法直接得到聚焦点。实验结果表明:所提出的方法可以实现准确、实时的自动聚焦。

Auto-focus is the important module of tool presetting and measuring machine based on computer vision,in which the definition evaluation function and focal point searching strategy directly affect the focus speed and precision.By comparing the calculation time,sensitivity,and single peak of the typical definition evaluation functions in this paper,the Brenner function and wavelet function are the optimal evaluation functions of the tool presetting and measuring system.To overcome the mechanical transmission error in focus search motion and improve the searching speed and precision,second-order discrete difference prediction model is proposed to predict the movement direction of the lens and direct the lens with large step quickly to the neighborhood region of the best-focused.Then,with narrow step length,a curve fitting method is used to directly get focus.Experimental results show that the proposed method can achieve accurate and real-time automatic focusing.