A Binary Document Image High Compression Ratio Compression Algorithm

As the intersection of binary images and the document images, and the subject part of binary images, binary document images have wide range of applications in the fax, e-governance, digital library construction and online marking etc.. And grayscale document images can be decomposed into binary document image by bit plane method. When it comes to huge amount of image data, the research on binary document image compression, not only can reduce the image storage space and the resulting costs,but also can reduce the access, processing and transfer bandwidth burden of the system, so it is essential to research the compression of binary document image.In this paper, the research achievements mainly include the following two aspects:1) This paper has done an in-depth study on layout analysis, and innovation on the key technologies. Layout analysis is very important step for binary document image compression, through classifying each region of document image, the document image compression ratio can be effectively improved. This paper has done innovation on the key technologies of layout analysis, and make a Matlab simulation. For commonly used method Otsu binarization, this paper give an improved method M-Otsu to obtain a better binary document image. This paper proposes a new skew detection method, the method can effectively improve the speed of skew testing and precision. Finally, use the mathematical morphology method to achieve the layout decomposition of binary document image.2) Entropy coding is improved, and implement a binary document image transform coding, transform coding ideas is introduced to binary document image compression successfully. In this paper, entropy coding method is improved by proposing a entropy coding method based on hierarchical strategy, which first of all to obtain the original image thumbnails through the contraction, and then a combination image of the corresponding pixels for the prospects pixels in the thumbnail. By this method the number of run-length in document image can effectively reduce. Another innovation of this paper is to achieve a binary document image transform coding. First, this paper gives a simple method of filter design and gives a fast algorithm for filtering, the experiment has given the multi-resolution wavelet transform decomposition and reconstruction of binary document image. After the experiment we can see that after the wavelet transform, the redundant statistical information had reduced, so less bits can be used in the compression of binary document image. Finally, on the basis of above work, a binary document image transform coding is achieved, the experiment results are good, so successfully introduce transform coding method into the binary document image compression.