人民币纸币冠字号码字符识别方法研究

以倾斜校正算法和模式识别算法2项核心技术为目标,充分利用霍夫变换法和最小二乘法在直线检测中的优点,采用一种结合霍夫变换法和最小二乘法的直线检测算法来求得人民币纸币图像的边缘直线及边缘直线的倾斜角度参数,并采用图像旋转变换算法实现对原倾斜人民币图像的倾斜校正。采用Hopfield神经网络的联想记忆功能,缩小待识模式与标准模板的差异,再用模板匹配法对Hopfield神经网络的输出结果进行识别,得到最终的识别结果。实验证明,结合Hopfield神经网络和模板匹配法的人民币冠字号码识别方法具备较高的识别率。

Study on recognition method of crown word number for RMB paper money

Slant correction method and pattern recognition algorithm are two key techniques in the RMB crown word number automatic recognition technology.According to the advantages of Hough transform method and the least squares method in line detection,a method combining Hough transform and the least squares is proposed to detect the edge line of the paper money image and find the angle parameter of the edge line.Image rotation algorithm is used to rotate the slanted image to get the correction image.Hopfield neural network,based on its function of association and memory,is designed to reduce the difference between the unknown pattern and the standard pattern.Pattern match method is programmed to recognize the output of the Hopfield neural network.Experiments show that RMB crown word number automatic recognition technology,based on Hopfield neural network and pattern match algorithm,has high recognition rate.