摘要
根据纸币CIS图像特点,先采用模糊集理论对纸币进行初次增强,提高目标区域对比度,然后扫描纸币边缘像素,采用最小二乘法拟合得到纸币亚像素边缘,拟合过程中,采用误差阈值法剔除噪声点,根据边缘角度对纸币进行倾斜校正并提取纸币区域,最后对提取后图像再一次进行模糊增强。实验表明,该方法通用性好,能有效地提高图像特征区域的对比度,在边缘磨损和缺角等情况下都能有效地提取纸币区域,并且增强后的图像有利于后续的纸币识别。
According to the paper currency CIS image＇s features,the fuzzy set theory is initially used to enhance the banknote target＇s contrast,then the paper money＇s edge pixels are obtained by scanning,they are fitted by least square method to get sub-pixel edge,in the fitting process,the error threshold method is applied for eliminating noises,then slope correction of paper money is carried out according to the angle of sub-pixel edge line and the note area is extracted,finally,fuzzy enhancement is used for the extracted image again.Experiments show that,the method has good versatility,it can enhance the contrast of bank note image effectively,it can even extract the currency area precisely in the edge deterioration and unfilled comer condition,and the enhanced image is in favor of subsequent paper money recognition.
出处
《科学技术与工程》
北大核心
2013年第34期10381-10385,10391,共6页
Science Technology and Engineering
