基于DSP的版面分析和OCR识别的原理及实现

The Theory and Implementation of Layout Analysis and OCR Recognition Based on DSP

随着光学字符识别 (OCR)技术广泛应用于人口普查、银行票据、各类报表等表格文档自动处理中 ,大大提高了表格文档的处理效率和精度 ,节省了人力物力和财力。 OCR识别系统通常采用扫描仪获取表格文档的原始图像 ,由计算机进行版面分析和字符识别。这里提出了基于 DSP的 OCR识别仪 ,采用接触式图像传感器摄取文档的原始图像 ,数字信号处理器(DSP)实现版面分析和字符识别。将识别结果输入到计算机中进行后处理。这样一台计算机可以联结多台识别仪 ,大大降低了 OCR识别系统的成本 。

Along with the widely use of the optical character recognition (OCR) in form document processing such as automatic census, bank cheques and report forms, not only a great amount of labor and money is saved, but also the processing efficiency and accuracy are greatly improved. The scanner is commonly used in OCR recognition system to acquire document original image, and the image transmit to computer for document layout analysis and OCR recognition. In this paper, OCR recognition instrument based on the digital signal processor (DSP) is put forward, and the contact image sensor (CIS) is applied to acquire the document image signal. The DSP is used for document layout analysis and OCR recognition. The processing result of DSP transmitted to computer for post processing. It is great significant in massive form document processing of that one computer can process the results of several OCR instruments.