票据识别系统数据获取过程研究  

The Research of CRS Data Obtaining Procedure

作　　者：庞韶宁[1] 李介谷[1] 

机构地区：[1]上海交通大学图象处理与模式识别研究所,上海200030

出　　处：《计算机工程》1997年第S1期287-289,375,共4页Computer Engineering

摘　　要：在手写体票据识别系统研究过程中,从带表格及噪声的背景中获取数据字符是至关重要的环节。实践证明,阈值图象分割方法是一种简单而有效的数据获取方法,该文以国家统一发票为研究对象,在样本直方图分析的基础上,讨论了阈值选取和样本明暗度之间的关系,并由此设计了一种新的票据获取过程结构。与通常方法比较,这使系统整体速度有了明显的提高。During the study of check pattern Recognition, to obtain the check data which is in the chart and noise background is key procedure of the CRS(Check Recognition System). And the threshold method usually is a simple and valid way to segment object image. TakingChina common check digital images as our study object, this article design a new procedure about the check data obtaining which is based on the analysis of the histogram of the check image. Used in practical CRS development, the result of higher speed of CRS is obvious.

关 键 词：明暗度 采样黑度 直方圆 阈值 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]