基于图像特征的汉字字形相似度计算
The Similarity Calculation of Chinese Character Glyph Based on Scale-Invariant Feature Transform
王昭 1杨婧 1杨敏1
作者信息
- 1. 国家计算机网络应急技术处理协调中心山西分中心,山西 太原 030012
- 折叠
摘要
汉字作为一种数量众多,结构复杂的象形文字,其字形的相似度计算对汉语的错别字识别和纠正具有重要意义.将汉字视为图形,通过尺度不变特征变换算法提取汉字的特征点,并利用汉字间特征点的对应关系,提出了一种基于特征点的相似度计算方法.对《通用规范汉字表》的8105 个汉字进行了相似度分析,结果表明,该方法和人的认知结果比较一致.
Abstract
As a pictograph with large number and complex structure,the similarity calculation of Chinese char-acters is of great significance for the recognition and correction of misspellings in Chinese.In this paper,Chinese characters are treated as graphs,feature points of Chinese characters are extracted by Scale Invariant-Feature Trans-form(SIFT)algorithm,and a similarity calculation method based on feature points is proposed by using the corre-sponding relationship between feature points between characters.The similarity analysis is carried out for 8105 Chi-nese characters and the results show that this method is more consistent with the human cognitive results.
关键词
汉字字形/形似字/相似度计算/尺度不变特征变换Key words
Chinese character glyph/being characters similar in form/similarity calculation/scale-invariant feature transform引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据