Scanning copy image stain removal method based on deep learning

Abstract

The invention discloses a scanning copy image stain removal method based on deep learning, and the method comprises the steps: firstly improving the traditional contrast enhancement through a deep learning contrast classification result, and improving the contrast of a scanning copy; then, marking main components of the scanned copy, performing deep learning training to obtain a model, detecting to obtain the main components of the scanned copy, and according to the main components, performing image subtraction to obtain stains at edge positions; further, on the premise that page number interference is eliminated by using deep learning, detection of point-like stain type stains is completed; and finally, automatically filling a stain area by analyzing color distribution around the stain. The system comprises an image preprocessing module, a contrast classification offline training module, a principal component detection offline training module, a dotted stain and page number detection offline classification module and a stain online detection module. According to the method, principal component detection and page number detection are utilized, and a deep learning technology and contrast improvement are combined, so that higher precision is obtained compared with the existing algorithm.

Description

technical field [0001] The invention belongs to the field of computer vision and pattern recognition, image processing and image stain removal systems, and in particular relates to a method for removing stains from scanned images based on deep learning. Background technique [0002] The digitization of paper archives has many advantages, such as long-term preservation of archives and easy retrieval. However, actual archival materials usually have some smudges. The existence of these smudges greatly affects the image quality of the scanned image, affects the image's beauty, readability, and even affects the subsequent OCR recognition and other operations. [0003] At present, there are relatively few such algorithms. The reason may be that the removal of smudges will slightly affect the authenticity of the scanned document (but there are many scenes that really need to remove smudges in the image), and it is more likely that the smudges themselves have irregularities. Comple...

AI Technical Summary

Problems solved by technology

[0008] The purpose of the embodiments of the present invention is to provide a deep learning-based scanning image stain removal method, which aims to solve the problem that the recognition accuracy of existing image stains is not up to standard and / or unsatisfactory, and the robustness is not strong The problem

Images