本文主要针对手写体数字串和定位格(线) 相粘连的情况,首次提出完整提取这种数字串的方法。首先运用数学形态学运算进行粗处理,去除定位格,得到特征点,然后结合数字的结构特征对它修补,最后进行平滑处理。实验表明本文方法的有效性。
Abstract
Until now , there is no method which can effectively ext ract characters overlapping position lines ( i. e , preprinted f rames) . This paper considers the case that the characters are handwritten numerals , and proposes a new ext raction method. This method consist s of two steps - coarse processing and fine processing. The first step removes the position lines in order to obtain st ructure feature point s by using mathematical morphology operation.The second step reconst ruct s broken character by considering numeral st ructure features ,and removes redundant st rokes through some st roke smoothing algorithm ,At last ,the above method is applied to the recognition problem of Chinese bank check.
关键词
数字串提取 /
手写体数字 /
数学形态学 /
字符重建
{{custom_keyword}} /
Key words
numeral st ring ext raction /
handwritten numerals /
mathematical morphology /
character reconst ruction
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1 ] 盛安东,师春礼,邮政编码自动识别系统的研制,中文信息学报,Vo1. 9 ,No. 3 ,1995 ,pp. 43 - 571
[2 ] S. Naoi ,M. Yabuki ,A. Askawa ,Y. Hotta ,Global interpolation in the segmentation of handwritten characters overlapping a border , IEICE Trans. Inf . & Syst . Vo1. E78 - D ,No. 7 ,1995 ,pp. 909 - 916.
[3 ]J . Serra , Introduction to Mathematical Morphology ,Vo1 ,CVGIP 35 ,1986 ,pp. 283 - 305.
[4 ] R. M. Harlick ,S. R. Sterberg ,X. Zhang , Image analsis using mathematical morphology , IEEE Trans ,Pattern Anal. Mach , Intell. ,Vo1. 9 ,No. 4 ,1987 ,pp. 532 - 550.
[ 5 ] L. Xu , E. Oja , Randomized hough transform ( RHT) : basic mechanisms , algorithms , and computational complexities ,CVGIP : Image Understanding ,Vo1. 57 ,No. 2 ,1993 ,pp. 131 - 1541
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(307 KB)
