Form content automatic entry method and device

Abstract

The invention provides a method and device for automatically typing in form content. The method includes the steps that (a), a scanning image of a form with content to be typed in is formed, and a scanning image of a blank form corresponding to the form with content to be typed in is determined; (b), a difference image of the two scanning images is obtained, and the initial positions of form entry whitespace areas are marked in the difference image; (c), the sum of pixel values of public external pixels of all the form entry whitespace areas in the difference image is calculated, and the matched position where the form entry whitespace areas are located when the sum of the pixel values is the minimum can be determined by horizontally moving and rotating the form entry whitespace areas in the difference image; (d), the form content in each form entry whitespace area at the matched position is recognized, and the recognition result is stored. By the method, interference and influence on form content recognition by form background and form content dislocation in the process of automatically typing in the form content can be avoided, and accordingly the accuracy of automatically typing in the form content is improved.

Description

technical field [0001] The invention relates to the technical field of image recognition, in particular to a method and device for automatically entering form content. Background technique [0002] At present, many units and institutions have to deal with a large number of forms every year and every month. In order to count and manage the contents of these tables, it is usually necessary to input the contents of these tables into the information management system first. [0003] Specifically, an invoice is taken as an example for description. Each type of invoice has a specific format. An invoice usually includes multiple items in predetermined positions, such as billing date, payment unit, invoice code, invoice number, industry category, machine-printed ticket number, query code, anti-counterfeiting code, item, amount, password area, total ( uppercase), total (lowercase), remarks, payee tax number, payee, biller, etc., on the right side or below each entry is the blank a...

AI Technical Summary

Problems solved by technology

However, the existing method of entering invoice content through automatic recognition has deficiencies in solving the following problems: 1) The recognition error rate of the existing Chinese recognition algorithm is relatively high, especially in the background interference (such as stamp residues, stains, invoices, etc.) 2) In many cases, the content of the invoice is printed out by a printer. When printing, the content of the printed invoice will be misplaced and deviate from its proper position (ie Deviate from the blank area it should fill in), thus increasing the difficulty of recognition and reducing the accuracy of recognition

Images