作　　者： (石伟）; (龙永红）; (钟云飞）; (杨丹君）; (周述庆）;

机构地区： 湖南工业大学电气与信息工程学院,湖南株洲412007

出　　处： 《包装学报》 2017年第1期66-71,共6页

摘　　要： 目前基于视觉的纸张计数方法大多是做图像分割或者边缘检测,虽然简单易行,但存在分割阈值难以选取的缺点。因此,提出一种基于灰度投影的硬板纸张计数算法。首先,将排列规律整齐的硬板纸张的图像通过灰度投影后得到其灰度像素曲线,再利用高斯函数得到的离散值作为权重,判断灰度像素曲线波峰和波谷的性质,然后根据波峰与波峰和波谷与波谷之间的间隔初步判断一张纸的厚度,并排除干扰峰谷点后准确定位出波峰或者波谷的位置,同时根据标记的位置实现对硬板纸张进行计数。最后,用实验验证了所提算法的有效性,实验计数结果表明,该方法能很好地实现对硬板纸的计数,且操作简单易行,2种不同硬板纸张计数结果的准确率均在98%以上。 At present,the paper counting methods based on vision are mostly connected with image segmentationor edge detection,which are simple,but there exists the difficulty in the choice of segmentation threshold.Therefore acardboard paper counting algorithm based on gray projection was designed.First of all,the arrangement orderliness ofcardboard paper image pixel gray curve was obtained through gray projection,and then Gaussian functions were used toget the discrete values as weights to determine the nature of the curve peaks and troughs.According to the interval betweenpeaks and between troughs,the initial thickness of a piece of paper was determined.By eliminating interference peak andvalley points,the position of the peak or valley was accurately located with the implementation of hard paper countingaccording to the position of the markers.At last,the experimental results showed that the algorithm could achieve effectivecounting,and the operation was simple and practical,with the accuracy rate of over98%with two different board paper.

关 键 词： 灰度投影 像素曲线 峰谷特征 纸张计数