Character recognition method and system for military scientific research and production data

Abstract

The embodiment of the invention provides a character recognition method for military scientific research and production data. The character recognition method comprises the steps of obtaining preprocessed military scientific research and production data, selecting a text region of an image by adopting a text detection technology, and outputting an initial recognition result and recognition accuracy of each character after character recognition is performed on the text region; obtaining an initial image of a text region corresponding to the extracted character, and based on a preset discrimination standard, discriminating the initial image through a preset discrimination module; if the discrimination result of the preset discrimination module is consistent with the initial recognition result, determining that the initial recognition result is correct; if the discrimination result of the preset discrimination module is inconsistent with the initial recognition result, determining that the initial recognition result is wrong; and determining the extraction probability of the extracted characters based on the document identification requirement, the character number, the identification accuracy of each character and the identification error frequency. In combination with document recognition requirements and character recognition accuracy, characters are randomly extracted for recognition, so that the proofreading efficiency is improved.

Description

technical field [0001] The invention relates to the field of data processing, in particular to a character recognition method and system for military scientific research and production data. Background technique [0002] In the long-term military scientific research and production process, a large number of paper documents have been accumulated, and it is urgent to rescue historical data, and use an automated method to quickly identify paper documents and form electronic data. Since military scientific research and production is a job that requires extremely high accuracy, the currently commonly used OCR (Optical Character Recognition, Optical Character Recognition) technology is difficult to meet the accuracy requirements. [0003] At present, OCR technology in the general field is relatively mature, such as Google OCR, Baidu OCR, Ali OCR, Hanwang OCR, etc., all provide convenient API interfaces, and users can directly call to realize the OCR recognition function. In addit...

AI Technical Summary

Problems solved by technology

[0004] At present, OCR technology is relatively mature, and the function of character recognition can be realized conveniently by calling the corresponding API interface. However, OCR technology only supports direct calling, and users cannot directly check the recognition accuracy of each character; at the same time, the existing technology recognizes accurately The rate is not high. For the field of military scientific research and production, after the machine is recognized, it is necessary to proofread the text again by professionals. Lack of data, it is difficult to improve the identification accuracy

Images