摘要
针对麻将包装在实际产线上出现的麻将牌错位、缺失、牌面翻转等问题,基于机器视觉检测技术设计了一套以DM6437处理器为硬件平台的麻将包装识别处理系统;该系统的硬件部分由摄像头板、核心板和接口板3个模块组成,配合机械传动系统来进行识别前后麻将包装的传输;系统的软件部分采用平板Pad对麻将包装进行模板录入,并设计了一套适应于不同机箱内部环境变化的背景剔除方法来完成模板整体分割,以及通过中值计算实现单目标的分离,最后还提出了一种简单高效的混合搜索方式来进行基于SAD的模板匹配;测试结果表明,识别算法对于麻将间有缝隙、麻将整体倾斜、个别或整体麻将翻转、个别麻将位置错乱等问题均能准确判别,模板匹配算法稳定性和适应性较好,识别准确率达95%以上,能满足包装流水线上的缺陷检测。
To resolve the problem of Mahjong Card misplacement,deficit,and reversal for Mahjong packaging production line,a visual processing system of Mahjong package recognition is designed based on machine vision and DM6437 processor.The hardware system contains the sensor board,DSP board and interface board,and coordinates with mechanical transmission system to transport the Mahjong package before and after the recognition.Meanwhile,the Pad is used to achieve the Mahjong package templates entering in the software system,and a background subtraction method adaptive to the environment changes of different mechanical systems is designed to complete the templates global segmentation.Besides,the separation of single objective is realized by median algorithm.Eventually,a simple and efficient mixed searching method is presented to achieve the template matching based on SAD.Experiment results on the recognition system show that the algorithm has a strong tolerability on the gaps between Mahjong cards,the tilt and reversal of Mahjong packages,and the misplacement of Mahjong cards.Furthermore,the stability and adaptability of the template matching algorithm are quite effective,and the accuracy of the identification achieves more than 95%,which can meet the detection quality on the packaging lines.
出处
《计算机测量与控制》
2016年第5期258-262,269,共6页
Computer Measurement &Control
基金
国家自然科学基金资助项目(61401398)
关键词
机器视觉
麻将包装识别
模板录入
模板整体分割
模板匹配算法
machine vision
Mahjong package recognition
templates entering
templates global segmentation
template matching algorithm