OCR切割性能自动评测  

AUTOMATIC PERFORMANCE EVALUATION OF OCR SEGMENTATION

作　　者：周开东[1] 许静[1] 

机构地区：[1]南开大学信息技术科学学院,天津300071

出　　处：《计算机应用与软件》2011年第7期105-107,共3页Computer Applications and Software

基　　金：天津市科技攻关项目(08ZCKFGX01100)

摘　　要：针对OCR(Optical Character Recognition光学字符识别)切割性能的评测问题,提出了一种基于切割图的评测方法,设计并实现了一个用于性能评测的自动化系统。该系统采用南开大学机器智能研究所自行研发的OCR图像切割处理核心RTK(Recog-nition Tool Kits识别工具包),根据切割样张集Cut-Set,计算切割错误统计数据、输出切割效果图像并对RTK进行修改。实验结果表明该评测方法显著提高了评测效率。Aiming at the performance evaluation of OCR（Optical Character Recognition） segmentation,in this paper,an evaluation method based on segmentation images is proposed,and an automatic system for performance evaluation is designed and implemented.The system adopts RTK（Recognition Tool Kits）,an OCR image segmentation processing kernel self-researched and developed by the Institute of machinery intelligence of Nankai University,and according to the sample Cut-Set,it calculates the statistical data of segmentation errors,exports segmentation effect images and modifies the RTK.Experimental results demonstrate the significant improvement in evaluation efficiency of the method.

关 键 词：性能评测 切割 OCR 识别率 RTK 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]