File image compressing method based on file image content analyzing and characteristic extracting

Abstract

The invention is a document picture compressing method based on analysis and characteristic extraction of the content of a document picture, comprising the steps of preprocessing document picture, cutting document picture, compressing character and compressing picture; where the first step accounts the gray value of the document picture and projects to picture boundary, and according to the edge variation of a projection curve and block diagram of gray values, analyzing the content of the document picture and automatically detecting and extracting the characteristic information of the document picture, including character height, picture boundary, attached mark position, pixel gray value, etc.; the second step cuts the document picture into character and picture parts according to the preprocessed result but the color information of the document picture is not influenced; the third step makes gray transform and travel coding on the character part for compressing; the last step makes the damaging compression on the picture part based on discrete cosine transform (DCT).

Description

technical field [0001] The invention relates to a document image compression method, in particular to a document image compression method based on document image content analysis and feature extraction. Background technique [0002] With the development of Internet and digital storage technology, as a substitute for paper documents, document images have been widely used in digital libraries, e-banking, e-government and other industries. At present, some websites already provide digital books, but generally they are scanned paper documents into images, and only a few of them use standard algorithms such as JPEG and JPEG2000 to compress the scanned images, but these algorithms are used for all parts of the image. The same compression technology, this compression method does not achieve good results for document image compression. At the same time, given that text strokes are oriented to human vision, compared with image distortion, human eyes are more likely to detect blur an...

AI Technical Summary

Problems solved by technology

Zhu Qingsheng, Lin Jie, and Zhang Min proposed a document image compression method based on layer segmentation in "Layer Segmentation-Based Document Image Compression", Computer Engineering and Design. 2004, Vol.25 No.8. It also does not consider the content and structural features of the document image, but uses multi-scale 2-color clustering to segment the document image layer, divides the image into foreground image layer, background image layer and mark image layer and compresses them respectively. This method needs to calculate the gray value of each pixel to determine which layer it belongs to, which requires a lot of calculations. When the image is segmented, it is divided into blocks of different sizes multiple times, and multiple iterative operations are also required; the method is more complicated. The method operation process is not intuitive enough

Images