首页> 美国政府科技报告 >Methods for Evaluating the Performance of Systems Intended to Recognize Characters from Image Data Scanned from Forms.
【24h】

Methods for Evaluating the Performance of Systems Intended to Recognize Characters from Image Data Scanned from Forms.

机译:用于评估从表格扫描的图像数据中识别字符的系统的性能的方法。

获取原文

摘要

The concepts presented in the paper were developed to establish a uniform method of evaluating the recognition of optical character readers used to process the information on forms which receive information as a bit stream directly or indirectly from scanners. Many large data entry systems are being designed to collect data from specified areas of forms, some of which may be multipart and completed with machine-printed or hand-printed characters. As the technology continues to advance, the number of commercially available products is increasing. Multiple products are emerging, all of which are designed to solve optical character recognition (OCR) applications. Improved recognition algorithms have enabled the accuracy of these products to steadily increase, but each product is based on a different, often proprietary set of algorithms. This presents potential users of OCR technology with many different choices and options and leads to a series of significant questions: How does a person determine when the technology has matured enough to make it economically advantageous to deploy; How does a potential user determine which product is best for his or her specific needs; How can a system developer, who has the ability to choose from a large variety of diverse algorithmic approaches, intelligently choose and then track progress when developing OCR systems. The answer to these questions lies in objective system performance measurement. This is the motivation behind the paper.

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号