image compression

Compressing compound images with a single algorithm that simultaneously meets the requirements for text, image and graphics has been elusive and thus requires new algorithms that can competently reduce the file size without degrading the quality. The compound image compression performance basically depends on the segmentation result. A segmentation process is used where regions of similar data types are grouped together. After successful segmentation existing techniques that best suits each data type can be used to achieve best compression results [4]. Segmentation algorithms for compound image compression are normally categorized as block based, object based and layer based segmentation. Each and every method has its own merits and demerits. Most of the recent researches in this field are mainly based on either layer based or block based. In object based method a page is divided into regions, where each region follows exact object boundaries. An object may be a photograph, a graphical object, a letter, etc. The main drawback of this method is its complexity. In layer-based method, a page is divided into rectangular layers [1]. Most layered coding algorithms use the standard three layer mixed raster content representation [5]. Some traditional compressors like DjVu [6], Digipaper [7] and JPEG 2000 [8] are also available.
In block-based method a page is divided into rectangular blocks where each block follows exact object boundaries. The advantages of this approach are simplified segmentation, better match between region boundaries and the compression algorithms, and the lack of redundancy [1]. Block-based approaches for compound images are also studied for their low complexity [5]. Block-based approaches segment non-overlapping blocks of pixels into different classes, and compress each class differently according to its characteristics [10]. The coding structures of block-based approaches are very flexible. Said et al. [9] proposed a simple blocked-based