烫印在线检测图像识别算法的研究  

Research of Image Recognition Algorithm for Hot Stamping Online Detection

作　　者：张冬娟[1] 唐万有[1] 朱捷航[1] 

机构地区：[1]天津科技大学,天津300222

出　　处：《包装工程》2013年第5期10-13,共4页Packaging Engineering

基　　金：广东省重大科技专项(2010A080402010)

摘　　要：应用中值滤波法对采集的烫印图像进行了噪声消除,根据烫印图案亮度值较高的特性,将彩色图像从RGB空间转换到了HSV空间,然后经过H通道阈值分割,结合差分求和定理以及模板分块匹配,改进了NCC算法,实现了烫印区域识别提取。通过大量实例对算法进行了验证,结果证明此算法能够快速、有效识别烫印区域,对机器视觉在烫印质量检测中的应用具有一定的指导意义。median filter method was applied to eliminate noise of collected hot stamping images. The color images were converted from RGB space into HSV space. H channel threshold was segmented. NCC algorithm was improved combining difference sum theorem and block template matching. Hot stamping area identification and extraction was realized. The algorithm was validated with a large number of examples. The results showed that the algorithm can quickly and effectively identify hot stamping area, which is significant for application of machine vision in hot stamping quality inspection.

关 键 词：烫印 图像识别 阈值分割 NCC模板匹配 

分 类 号：TS807[轻工技术与工程] TS851.6