应用机器视觉的鞋面数量统计与尺寸测量

发布时间：2018-04-18 15:23

本文选题：鞋面数量 + 鞋面尺寸　；参考：《纺织学报》2017年11期

【摘要】：针对鞋面数量统计和尺寸测量过程中人工检测方法效率低、成本高的问题,提出利用机器视觉的鞋面数量统计与尺寸测量方法。采用基于边缘方向的模板匹配算法对鞋面进行数量统计。运用最小外接矩形(MER)算法对鞋面进行尺寸测量。为验证方法的可靠性,对3种鞋面分别在光照变化、混乱和遮挡条件下进行了检测实验,并与绝对误差和算法、归一化相关系数法和Hausdorff距离匹配算法的检测结果进行对比。将MER测量结果和人工测量结果进行对比,结果表明:基于边缘方向的模板匹配算法在复杂环境下的鲁棒性更高,可完成鞋面数量统计;MER方法和人工测量的相对误差在0.67%以内,满足工业生产中的实际需求。
[Abstract]:In order to solve the problems of low efficiency and high cost in the process of number statistics and dimension measurement of upper, a new method based on machine vision is put forward.A template matching algorithm based on edge direction is used to calculate the number of shoe upper.The minimum outer rectangle Mer algorithm is used to measure the size of the upper.In order to verify the reliability of the method, three kinds of uppers were tested under the condition of illumination variation, confusion and occlusion, and the results were compared with those of the absolute error sum algorithm, the normalized correlation coefficient method and the Hausdorff distance matching algorithm.Comparing the results of MER and manual measurement, the results show that the template matching algorithm based on edge direction is more robust in complex environment, and the relative error of Mer method and manual measurement is less than 0.67%.To meet the actual needs in industrial production.
【作者单位】：西安工程大学电子信息学院;
【基金】：国家自然科学基金项目(61301276) 陕西省重点研发计划项目(2017GY-003) 陕西省工业科技攻关项目(2015GY034) 西安工程大学研究生创新基金项目(CX201730)
【分类号】：TP391.41;TS943.79

【相似文献】