Adaptive processing and archiving of compound scanned documents

摘要

In the paper is presented one new approach for adaptive processing and compression of images of scanned documents, which contain text and pictures. In order to achieve high compression with maximum retained quality, the document content is analyzed and two corresponding regions of interest are set. Then, each region is processed as follows: the text - with lossless and the picture - with lossy compression. The lossy compression is based on the Inverse Pyramid Decomposition, and for the lossless compression is used the adaptive run-length coding method. The quality of the pictures in the restored document is visually lossless, while the text is unchanged. The overall compression efficiency surpasses that of the JPEG 2000 standard for same images. In the paper are presented the basic principles of the Inverse pyramid decomposition and are included some experimental results and comparison with JPEG 2000.