基于H-KNN的藏文字符的识别研究  

Research on the Tibetan Characters Recognition Based on H-KNN

作　　者：吴玉龙 卓嘎[1] 扎西平措[1] 赵智龙 吴绍乾 WU Yulong;ZHUO Ga;ZHAXI Pingcuo;ZHAO Zhilong;WU Shaoqian(Tibet University,Lhasa 850000,China)

机构地区：[1]西藏大学,西藏拉萨850000

出　　处：《现代信息科技》2022年第8期92-94,共3页Modern Information Technology

基　　金：西藏自治区级大学生创新训练项目(S202110694079)。

摘　　要：在OCR技术越来越成熟的今天,中文OCR技术早已发展成熟,但是藏族聚集地的藏族同胞所使用的藏文OCR技术却还未成熟。针对此,文章通过改进KNN算法,增加希尔伯特曲线来改进算法,设计了基于H-KNN的藏文字符识别的算法,利用最近邻算法与希尔伯特曲线相结合的方法来识别藏文数字字符,改进了字符在预处理时的降维方式,提高了KNN算法的识别效果,实验结果证明,相较于传统的KNN算法识别正确率有显著提升。Today, the OCR technology is becoming more and more mature, and Chinese OCR technology has long been developed and matured, but the Tibetan OCR technology used by Tibetan compatriots from Tibetan gathering areas is not yet mature. Aiming at the situation, this paper designs a Tibetan character recognition algorithm based on H-KNN through improving the KNN algorithm and increasing the Hilbert Curve to improve the algorithm. It uses the method of combining the KNN algorithm with Hilbert Curve to identify Tibetan numeric characters, improves the dimensionality reduction mode of characters in preprocessing, and improves the recognition effect of KNN algorithm. The experimental results prove that there is a significant improvement in the identification accuracy rate compared with the traditional KNN algorithm.

关 键 词：藏文字符识别 最邻近算法 希尔伯特曲线 OCR 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]