Journal of Computer Applications ›› 2013, Vol. 33 ›› Issue (03): 645-647.DOI: 10.3724/SP.J.1087.2013.00645

• Network and distributed techno • Previous Articles     Next Articles

Correlation matching recognition algorithm based on character normalized double projection

WAN Jin'e1,2*, YUAN Baoshe1,2, GU Zhao3, Mirsali SABIT1,2   

  1. 1.School of Information Science and Engineering, Xinjiang University, Xinjiang Urumqi 830046, China;
    2.Key Laboratory of Multilingual Information Technology, Xinjiang University, Xinjiang Urumqi 830046, China;
    3.School of Mathematics and System Science, Xinjiang University, Xinjiang Urumqi 830046, China
  • Received:2012-09-28 Revised:2012-12-12 Online:2013-03-01 Published:2013-03-01

基于字符归一化双投影互相关性匹配识别算法

万金娥1,2*,袁保社1,2,谷朝3,米尔沙力江·沙吾提1,2   

  1. 1.新疆大学 信息科学与工程学院,乌鲁木齐 830046;
    2.新疆大学 新疆多语种信息技术重点实验室,乌鲁木齐 830046;
    3.新疆大学 数学与系统科学学院,乌鲁木齐 830046
  • 通讯作者: 万金娥
  • 作者简介:万金娥(1987-),女,甘肃玉门人,硕士研究生,主要研究方向:图形图像识别处理; 袁保社(1955-),男,新疆乌鲁木齐人,教授,主要研究方向:图形图像处理、电路理论、信号处理; 谷朝(1987-),男,山东菏泽人,硕士研究生,主要研究方向:图论、组合数学; 米尔沙力江·沙吾提(1985-),男(维吾尔族),新疆喀什人,硕士研究生,主要研究方向:图形图像识别处理。
  • 基金资助:

    工业和信息化部电子信息产业发展基金资助项目(工信部财[2009]453)。

Abstract: Since the correct character recognition rate of the printed Uighur character recognition system is low, in this study, a character image was scanned in both horizontal and vertical directions to create a row-projection vector and a column-projection map. In combination with three levels of classification, the target character with the corresponding classification characters' double projection maps was normalized individually, and the correlation mean calculation was carried out. The maximum mean character was taken as the best match recognition result, so as to implement the Uighur character recognition. The correlation matching recognition algorithm based on character normalized double projection method has been proved by experiment that it has the advantages of strong anti-interference, simple implementation, high matching accuracy, thus improving the correct rate of the printed Uighur character recognition.

Key words: printed Uyghur, double integral projection map, projection map normalization, mutual correlation, template matching recognition

摘要: 针对印刷体维吾尔文文字识别系统中的字符识别正确率较低这一难点问题,采用对字符图像进行横向扫描和纵向扫描生成行和列投影图, 结合三级分类,将目标字符与对应分类中的字符的双投影图逐一归一化并进行相关性均值计算的方法,取均值最大的字符作为最佳匹配识别结果,实现了对维文字符的识别。实验证明这种基于字符归一化双投影互相关性匹配识别算法方法抗干扰性强,简单易行,匹配精度高,使得印刷体维吾尔文字字符识别的正确率有了进一步提高。

关键词: 印刷体维吾尔文, 双积分投影图, 投影图归一化, 互相关性, 模板匹配识别

CLC Number: