CAJ | 학술논문

针对铁路客票自动识别研究中票号与背景难以分离的票号分割问题,提出一种基于改进的离散KL变换(DKLT)的彩色铁路客票票号分割方法.首先,将彩色客票图像进行DKLT,然后,选择主要的特征信息进行图像重构,并采用三维空间直方图来检验处理后票号在图像中的聚类效果来滤除背景,最后,辅助以合适的形态学运算进行票号的有效分割,为铁路客票的识别提供条件,实验结果表明了该方法在铁路客票票号分割应用上的可行性.
침대철로객표자동식별연구중표호여배경난이분리적표호분할문제,제출일충기우개진적리산KL변환(DKLT)적채색철로객표표호분할방법.수선,장채색객표도상진행DKLT,연후,선택주요적특정신식진행도상중구,병채용삼유공간직방도래검험처리후표호재도상중적취류효과래려제배경,최후,보조이합괄적형태학운산진행표호적유효분할,위철로객표적식별제공조건,실험결과표명료해방법재철로객표표호분할응용상적가행성.
Aimed at the problem of railway passenger tickets automation recognition, the codes segmentation method based on discrete Karhunen-Loeve transform(DKLT) algorithm is presented. Firstly ,the railway passenger tickets color image is implemented DKLT. Then the corrected character information is selected in the reconstruction to disposal the background. The 3D histogram is used to see the ability of clustering. At last, the mathematical morphology is integrated in the course of the railway tickets cedes segmentation. Those give convenience to the tickets recognition. The experiment results show the feasible application of the method given above in the tickets codes segmentation.