文档图象中特定信息的智能识别

本文讨论了一种基于光学标记识别技术的文档图象识别系统，重点阐述了该系统的设计思想和实现技术。系统以标记识别为例，对图象数字化、图象预处理、获取图象信息以及信息特征提取等几个关键部分进行了详细描述。在图象预处理过程中给出了一种高效实用的边缘检测算法，在标记信息特征提取中采用了统计分析的方法，其结果可以大大
大提高标记识别系统的准确率。

Intelligent Recognition of Specific Information in Document Images

This paper discusses a document image analysis system based on optical mark recognition. The design method and implementation techniques are emphasized. We use mark recognition as an example to describe the key components of the system, image digitalization, preprocessing of images, image information acquisition and feature extraction. A highly practical and efficient boundary detection algorithm is presented in preprocessing. Statistical analysis is applied to mark feature extraction. These methods greatly enhance the recognition rate.