摘要
针对尺寸法进行人民币的面额识别中,使用Freeman链码进行纸币边界曲线跟踪时出现的断点、漏点、错点及处理数据量大、速度慢的问题,提出一种新的链码定位边界法。首先通过统计纸币轮廓点的横、纵坐标值出现的频率确定链码起始点;然后定义一种新的多方向链码以解决图像边界点的间断问题;最后利用之前链码预测之后链码以减少漏点、错点,减少噪声对曲线跟踪的影响。实验结果表明,该方法用于纸币识别,识别率达到了96.159%以上,计算复杂度降低,识别速度提高,是一种有效的图像边界提取方法。
For the process of currency denomination identification which uses of the size identification method, occurs with the problem of breakpoints, leakage points, fault points and large volumes of data when using Freeman chain code curve track the notes boundary, a new chain code method is presented for boundary locating. First of all, the starting point has been determined by statistical the frequency of the horizontal and vertical coordinates of contour points of notes, and then a new muti-directional chain code is defined to solve the image boundary points＇ intermittent problem, finally, chain code prior is used to forecast the chain code to reduce leakage points and wrong points, reduce the impact of noise on the curve tracking. The experimental results show that the method used for banknotes recognition, the recognition rate is more than 96. 159%, the computational complexity has been reduced, the recognition speed has been increased and is an effective image boundary extraction method.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第12期4643-4646,共4页
Computer Engineering and Design
关键词
面额识别
尺寸识别法
纸币定位
曲线跟踪
FREEMAN链码
denomination identification
size identification method
banknotes positioning
curve tracing
Freeman chain code
