Ancient document image stain removal method

Abstract

The invention discloses an ancient document image stain removal method. The method comprises the following steps of: converting a to-be-processed image from an RGB color space to a Lab color space, separating the to-be-processed image by utilizing three channel images of the Lab color space, and selecting an L channel and a b channel by utilizing own information of separated different channel images so as to weaken or eliminate the influences of stains; automatically determining image text blocks according to the quantity and sizes of characters in the stain eliminated image, and judging whether the blocked image needs expansion or not; and carrying out global and local combined binarization processing on the image blocks to obtain a binary image. According to the method, the three channelimages of the Lab color space are separated and fusion is carried out on two channels, so that the problem of removing stains in ancient document images is solved; and during the binarization processing, the global and local are combined by adoption of an automatic image blocking method, so that the stains in the ancient document images can be effectively removed and the binarization effect is good.

Description

technical field [0001] The invention belongs to the technical field of document analysis and recognition, and in particular relates to a method for removing stains from images of ancient book documents. Background technique [0002] Image binarization is often used in document image restoration. Image binarization can be divided into grayscale image binarization and color image binarization. Therefore, the selection of the threshold has a decisive effect on the image binarization result. Due to the age, many pages of ancient books have stains. How to remove the stains, obtain effective binary images of Tibetan ancient books, and then perform document segmentation and character recognition is an important link. [0003] At present, the threshold-based binarization method can be divided into global threshold method and local threshold method. The global threshold method uses a single threshold for the entire image, and judges the target and background by comparing all pixel v...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the problems in the prior art that the stains in the image cannot be removed by methods such as Otsu, Niblack, and Sauvola, and the binarization effect is poor, the present invention provides a method for removing stains from images of ancient book documents, which uses the Lab color space Three channels, to obtain the position information of the stain to weaken and eliminate the influence of the stain, and use the automatic block method to binarize the image blocks to obtain the binary image of the entire image, laying the foundation for further line word segmentation and recognition

Images