摘要
为了消除传输纸机的机械传动部分存在偏差而引起的钞票纸大张图像在空间位置上的偏移,以及因生产工艺波动而导致钞票纸的防伪特征区在大张图像中空间位置的波动,提出一种在钞票图像检测之前对大张图像和防伪特征区进行两级搜索定位的方法。该方法首先使用差平方和法对采集的大张图像四边进行搜索定位,然后针对不同的防伪特征区分别采用改进的快速相关函数法和基于核的直方图统计法进行搜索定位。实验结果表明该方法能够为后续的检测提供高效、精准、高质量的检测条件。
In order to eliminate spatial offset of large image of banknote paper which is caused by deviations ex- isting in the transfer machine＇ s mechanical transmission part, as well as spatial wave offset of security feature area in large image which is caused by fluctuations in production process, a two-level search positioning method of the large image and the security features before banknote image detection was proposed. This method firstly employs sum of squares method to locate edges of large image, and then uses improved rapid correlation function method and histogram statistics method based on kernel function to locate different security feature areas. The experimental re- sults show that the method can provide efficient, accurate and high-quality detection conditions for the follow-up banknote image detection.
出处
《科学技术与工程》
北大核心
2012年第26期6813-6816,共4页
Science Technology and Engineering
基金
河南省基础与前沿技术研究计划项目(102300410266)
(122300410287)
郑州轻工业学院博士科研基金(2010BSJJ038)资助
关键词
两级搜索定位方法
钞票图像检测
差平方和法
相关函数法
直方图统计法
two-level search positioning methodcorrelation function methodbanknote image detectionhistogram statistics methodsum of squares method
