Chinese business card OCR (optical character recognition) data correction system utilizing massive associated information of knowledge base

Abstract

The invention provides a Chinese business card OCR (optical character recognition) data correction system utilizing massive associated information of a knowledge base. The Chinese business card OCR data correction system comprises an image collection module, an image standardized processing module, a block extracting module, an OCR module, a knowledge base module, a data correction module, a gain maintaining module and a result displaying module. The system is characterized in that to-be-corrected data are labeled by subjecting recognition results of the OCR module to information structuralized processing; address and organization name associated information is corrected by utilizing the massive associated information of the knowledge base module and combing a series of techniques like Chinese word segmentation, importance weighting based on the knowledge base, similarity comparison based on texts and images and information integration to improve accuracy; corrected OCR results are output and displayed. In addition, the gain maintaining module of the system performs information maintaining on the knowledge base in a semiautomatic manner to meet needs of continually-growing of information quantity.

Description

technical field [0001] The invention belongs to the technical field of optical character recognition and the field of data cleaning, and in particular relates to a Chinese business card OCR data correction system based on massive associated information in a knowledge base. Background technique [0002] With the development of optical scanning, computer image processing, pattern recognition and other technologies, OCR technology has gradually matured, and its successful application in many aspects has also brought convenience to people's work and life, and business card OCR recognition is one of the most In a representative application, the user only needs to transfer the photo of the business card to the computer or smart phone, and use the corresponding OCR software to recognize it, and then store the recognized text information, which saves people from using the traditional Ways to deal with the troubles caused by business cards, such as the inconvenience of carrying busin...

AI Technical Summary

Problems solved by technology

[0003] However, the current OCR technology itself is affected by many unfavorable factors, such as low illumination, low pixels, image noise, angle tilt, aggregation blur, etc., resulting in a low final recognition rate

There are basically two ways to correct the OCR results. One is to start from the image itself and try to weaken or eliminate the influence of the adverse environment through image denoising, image enhancement and other technologies. This method only relies on image information to correct The error effect is limited; another is to correct a certain type of information such as the address separately. The disadvantage of this method is that it does not take advantage of the correlation of the data, so there is still room for further improvement in the correction effect. space

Images