Research On ROI Coding Of Scanned Document Image

Scanned document image becomes a main fundamental representation of information in office documents and digital libraries etc. In this paper, author presents a document image processing scheme, and mainly studies several key techniques in implementation. The main works are as follows:First, the paper states the features of document image, and the functional properties of a document image processing system. And some effective preprocessing can get a better visual effect of the compressed image. The main methods include removing noise from background and reinforcing the text.Second, this paper presents an image compression scheme based on ROI (Region of Interest). Author segments an image and then makes the text layer as ROI of document image, then he compresses and reconstructs its all layers according to its own characteristic. Author expects to obtain a higher compression rate than general image compression methods.Third, to compress the text layer and image layer respectively. The whole process of compression includes sub-sample, wavelet transform, EZW and arithmetic encoding. And the whole process of decompression includes arithmetic decoding, EZW decoding, wavelet reverse transform and image smoothing.Fourth, to make use of the method of progressive transmission. Transmission by twice can effectively relieves the burden of network.Fifth, through a post-processing according to the compress scheme, effect of reconstructed image is reinforced. The main methods include image smoothing and reinforcing reconstructed image.Finally, all of stated above are implemented in a demonstration software. It can be concluded that the document-image processing model combines the theories and the practices. Most of the ideas, methods, and implementation techniques discussed in this paper are directly applicable during the process of developing digital library and e-government based on network. The paper includes two main initiations. One is employing the method of segmentation before compression, which is benefit to a more effective compression according to the characteristics of scanned document image. Another is combining ROI and progressive transmission, in which makes the compression rate and transmission performance reach the best balance.