Imaging process for bank bills is vulnerable to the adverse effects for complex background interference or lack of illumination or uneven illumination,which makes the task of accurate auto-cropping for bank bills be very challenging in real-time.After automatic correcting of the deflection angle of the bills,a fast auto-cropping strategy based on adaptive im-age binarization is presented.The policy is applied to the integration testing with the auto-processing device for bank bills, results show that the policy has a good ability for the suppression to background interference and can obatin accurate cropping results in the environment with natural illumination,expecially when the background color is near to the base color of the bill.The overall correct rate is 95%,indicates that the self-adaptability and robustness of the methodology proposed is supe-rior to that of the method based on edge detection.In addition,the median filtering algorithm was improved,which greatly reduced the time complexity,so as to meet the need of the real-time online processing for bank bills.%银行票据图片采集过程易受到复杂背景干扰和光照不足或不均等不利影响,使得自动、准确、实时裁剪票据的任务很具挑战性.在票据自动纠偏基础上,提出了基于图像自适应二值化的银行票据自动、快速裁剪策略.将该策略应用到银行印控仪进行集成测试,不同成像背景和光照环境下的多组数据结果表明本策略在自然光照环境中,能够很好地抑制背景干扰,准确裁剪票据,而在背景颜色接近票据底色时或存在明显光斑干扰情况下,依旧能够获得正确结果,总体裁剪正确率达到95%,其自适应性和鲁棒性明显优于基于边缘提取的方法.且改进了快速中值滤波算法,大大降低了滤波处理时间复杂度,可满足银行在线处理的实时性需求.
展开▼
