Bill image position accurate correction method based on title position reference template

Abstract

Disclosed is a bill image position accurate correction method based on title position reference template, the title position reference template-based bill image position accurate correction method isrealized by taking the title position reference template as a reference object for a specific bill and automatically adjusting the direction and the position of the bill step by step. A user only needs to click the first character and the last character of the bill title in the bill image, the position, direction and size of the bill can be corrected more accurately and automatically, and a solidfoundation is laid for recognizing key content of the bill.

Description

Technical field: [0001] The invention relates to the technical field of image position calibration processing, in particular to a method for accurately correcting a bill image position based on a title position reference template. Background technique: [0002] In the current modern financial management process, many enterprises and units have electronically stored bills. If the specific context information in these bills can be identified (such as: a list of consumables purchased at one time, and the cost details of a specific hotel), corporate financial bills can be formed Content big data, these data can provide important data support for enterprise management, finding financial loopholes and building a clean government, so it is very necessary to identify the content of bill images. [0003] The current OCR technology is relatively mature and can easily recognize the text in the image. However, the content of the bill is different from the general book content, and the r...

AI Technical Summary

Problems solved by technology

This method can achieve a high level of automation. However, in actual work, it is difficult to strictly limit the bill shooting background and image cropping methods used by reimbursers. In some cases, there are multiple colors in the shooting environment background, and the shooting equipment is distorted at the image boundary. Different situations, improper cropping of the image file, and multiple orientations of the text inside the bill will all affect the correction process described above, making it difficult to achieve the goal of "accurate position" in the above process, and the final calibrated image position will be different. There are random errors, which ultimately lead to the inability to rigorously determine the desired region in the image

The second is to use a purely manual method to continuously observe and adjust the bills. Although this method can solve the complexity of the scanned image, it requires a lot of interactive and iterative operations between humans and computers, so it consumes time and labor costs. High, it is difficult to meet the requirements when the business volume is large

Images