Document positioning and cutting method in picture

Abstract

The invention discloses a document positioning and cutting method in a picture. The method comprises steps of S1: inputting a document image; S2: carrying out straight segment detection on the document image, and classifying the detected document segments into nearly horizontal straight segments, nearly vertical straight segments and straight segments in other directions; S3: extracting horizontal boundary segments and vertical boundary segments; S4: according to the extracted horizontal boundary segments and vertical boundary segments, determining positions of four top points of the document image; and S5: according to the positions of the four top points, carrying out cutting and righting on the document image. According to the invention, quite complex document types can be processed; top points can be quite precisely positioned in case of a quite complex background; the method is applied to software for document picture processing; after a user uses a mobile phone to shoot the document, the documents in the image can be quickly cut and righted; interference can be eliminated for following document identification modules; and identification rate of words in the documents is improved.

Description

technical field [0001] The invention relates to the technical field of computer software, in particular to a method for locating and cutting documents in pictures. Background technique [0002] In traditional office processes, the digitization of files or documents is usually done through scanners. With the popularization of smart phones and the improvement of the quality of mobile phone cameras, more and more consumers begin to use mobile phones to take pictures of documents or documents to obtain digital copies of documents. However, due to the limitation of the shooting angle and viewfinder range, the document pictures taken directly by the user often have the following two problems: (1) the document often only occupies the center part of the picture, and there are a lot of invalid background interference around the document, which needs to be cut off. (2) It is difficult for the user to keep the mobile phone parallel to the desktop where the document is placed when shoo...

AI Technical Summary

Problems solved by technology

However, due to the limitation of the shooting angle and viewfinder range, the document pictures taken directly by the user often have the following two problems: (1) The document often only occupies the center part of the picture, and there are a lot of invalid background interference around the document, which needs to be removed

(2) When the user is shooting, it is difficult to keep the mobile phone parallel to the desktop where the document is placed, causing the document to be tilted in the image, which needs to be corrected

[0004] In the existing software on the market, some of the four vertices of the document require the user to specify manually. This process requires user interaction, which is inefficient and not suitable for processing a large number of images.

There are also some software that automatically locate the four vertices of the document through image processing, but due to the limitations of the algorithm, misjudgments often occur, and the success rate is low

Images