Bill information processing method, system and computer-readable storage medium

Abstract

The invention discloses a bill information processing method, a system and a computer-readable storage medium, wherein, the bill information processing method comprises the following steps: a depth neural network model is used for training a bill of a specific type to obtain a positioning model of a plurality of target text regions on the bill of the specific type; A target text area on that billto be processed is located accord to the positioning model; Forming an attribute classification model by using machine learning and classifying the target text regions located on the bill to be processed to determine the attributes of at least one target text region; According to the attributes of the determined target text area, the relative positional relationship between the located target textareas and the attribute classification model, the attributes of the target text area whose attributes are not determined on the bill to be processed are obtained, and / or the attributes of the determined target text area on the bill to be processed are subjected to the first verification correction. The invention can effectively improve the accuracy rate of the attribute classification of the billinformation positioning.

Description

technical field [0001] The present invention relates to the technical field of image processing, in particular to a bill information processing method, system and computer-readable storage medium. Background technique [0002] In daily work or life, all kinds of bills are used. The checking and statistical work of bills is not only a large amount of business, but also requires professionals to complete. With the rapid application of computer technology in various fields, the processing of bill information is done by machines. It has also become a trend. [0003] At present, the machine mainly uses Optical Character Recognition (OCR) technology to extract information printed on paper, which uses optical technology and computer technology to read the text printed or written on paper and convert into a format that computers can accept and humans can understand. The processing steps of OCR mainly include: image preprocessing, layout analysis, text positioning (or image cutting...

AI Technical Summary

Problems solved by technology

This text positioning method can better deal with low-complexity layout information such as books, newspapers and periodicals. However, when dealing with high-complexity bill layout information, if the bill information is printed skewed or misplaced, and the text is underlined, positioning will occur. Inaccurate, resulting in the inconsistency between the located text attributes and the actual text attributes. Therefore, the error rate of bill information recognition and extraction is high, which makes machine processing impossible

Images