机器视觉系统的自动对焦技术研究  

Research on Auto-focusing Technology of Machine Vision System

作　　者：王灿 张红霞 刘鑫 白志城 付秀娟[1] 王刚 王学华[1] Wang Can;Zhang Hongxia;Liu Xin;Bai Zhicheng;Fu Xiujuan;Wang Gang;Wang Xuehua(Wuhan Institute of Technology,Wuhan430200,China;Wuhan Superscene Information Technology Co.,Ltd.,Wuhan430204,China)

机构地区：[1]武汉工程大学材料科学与工程学院,武汉430200 [2]纵横皆景(武汉)信息技术有限公司,武汉430204

出　　处：《计算机测量与控制》2019年第3期150-154,共5页Computer Measurement &Control

基　　金：国家自然科学基金(51575406);武汉工程大学研究生创新基金(CX2017019)

摘　　要：为了解决在工业生产现场复杂环境下机器视觉系统中定焦相机的失焦问题,提出了一种机器视觉定焦相机的自动对焦技术,该技术通过上位机实时对相机采集到的图像进行感兴趣区域(Region of Interest,ROI)裁剪、离散傅里叶变换(Distributed Fourier Transformation,DFT)滤波等预处理过程,再由改进的图像清晰度评价函数对图像清晰度进行评价,最后由运动机构带动相机自动搜寻焦点位置并完成对焦功能;实验结果表明,经过傅里叶变换滤波后,提升了评价函数的灵敏度;通过与几种典型的评价函数对比,提出的方案能有效地排除环境干扰,自动对焦的结果与标准样张最为接近,且在不同物距情况下仍能够高效运行。In order to solve the problem of fixed camera out-of-focus in the vision system in the complex environment of industrial production site, an auto-focusing technique for vision camera is presented, which can be used to preprocess ROI regional cutting, distributed Fourier transform, and other preprocessing by using the image captured by the host computer in real time. Then the image definition is evaluated by the improved definition estimation function of digital image, and the camera driven by motion mechanism automatically searches the focus position and completes the focusing function. The experimental results show that the sensitivity of the evaluation function is improved after Fourier transform filtering. Compared with several typical evaluation functions, the proposed scheme can effectively hold aside the environment interference, the result of auto-focusing is closest to the standard samples, and can be run efficiently under different object spacing.

关 键 词：机器视觉 自动对焦 运动控制 图像清晰度 

分 类 号：TB472[一般工业技术—工业设计]