摘要
在机器视觉检测中,图像光照不均匀现象会增加后续处理的难度,因此需要对其进行有效的阈值分割。算法通过窗口分割提取原图的背景灰度图后,结合局部对比度调整系数,对图像进行背景均匀化处理,然后进行全局阈值分割。实验对具有典型光照问题的高分辨率线纹尺图像处理效果良好,平均时间在0.5 s以内。通过与其他几种算法的对比,证明了本算法处理效果最佳,所耗时间满足实时性,为目标的进一步测量工作奠定了良好的基础。
In machine vision inspection, uneven illumination images will increa!~e the difficulty of the subsequent image processing. Therefore, uneven illumination images need effective threshold segmentation. After extracting the background grayscale of the original image through window division, this paper homogenized the background of the image combined with the local contrast adjustment coefficient. Then used global threshold segmentation to binarize the image. Images used in the experiment are linear scale images with typical lighting problems and a hign resolution. The resuhs indicate that the effect of the binarization is pretty fine and the average processing time is within 0.5s. Through comparison with several other algorithms, it demonstrates that this algorithm can ensure the best processing effect and the consumption of time can meet the requirement of real-time industrial measurement, so that provides a good foundation for further measurement
出处
《计算机应用研究》
CSCD
北大核心
2015年第11期3467-3470,共4页
Application Research of Computers
基金
国家重大科学仪器设备开发专项(2013YQ17053903)
关键词
机器视觉
光照不均匀
背景提取
阈值分割
machine vision
uneven illumination
background extraction
threshoid segmentation
作者简介
王仲(1953-),男,天津人,教授,硕士,主要研究方向为精密机械与机器视觉;
郑镕浩(1990-),女,山西吕梁人,硕士研究生,主要研究方向为机器视觉、图形图像处理;
付鲁华(1972-)(通信作者),女,山东人,副教授,博士,主要研究方向为精密测试技术及仪器(fuluhua@tju.edu.cn);
苟建松(1991-),男,山东滨州人,硕士研究生,主要研究方向为DSP自动控制.