摘要
为准确地定位出目的地址块 ,针对中文手写信封具有的明显特点 ,提出了一种用于中文手写信封的目的地址块定位方法。二值图像在经过噪声滤波等预处理后 ,提取出连通域。用模式分类的方法将非文本特性的连通域去除 ,并由文本块合并成文本行。在已提取的文本行基础上定位出目的地址块。对 5 0 0个采集于邮政分拣机的实际样本的提取率和平均 CPU时间分别约为 85 %和 3.7s。结果表明该方法是有效的并且花费较少的
Recognition of the destination address block (DAB) is needed for postal automation and the successful location of the DAB is a prerequisite to recognition of the DAB. This paper proposes a new method to locate the DAB on handwritten Chinese envelopes, using the obvious characteristics of handwritten Chinese envelopes. The components were extracted from a 2 bit image after preprocessing steps such as noise filters. Nontext and text components were classified and the latter were merged into text lines. The DAB was identified by analysis of the basis of text lines. The location rate and average CPU time for the 500 envelopes obtained from mail sorting machines are about 85% and 3.7s, respectively. The results show that the method is effective and takes short CPU time.
出处
《清华大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2000年第1期35-38,共4页
Journal of Tsinghua University(Science and Technology)
基金
国家自然科学基金项目!(6 96 82 0 0 3)
国家"八六三"高技术项目! (86 3- 30 6 - 0 3- 0 5 - 6 )
关键词
目的地址块
邮政自动化
中文手写信封
定位
destination address block (DAB)
postal automation
handwritten Chinese envelopes
text line extraction
