An improved prediction method for quadtree image coding

Quadtree segmentation is a simple technique to represent an image at different resolution levels. This representation can be used to compress an image efficiently; however, at the same time, it may cause sawteeth between sub-images. This paper proposes a better reconstruction algorithm for quadtree image coding. It can eliminate the sawteeth between sub-images and improve the quality of the reconstructed image. The algorithm of the proposed method is quite simple, so it would not decrease the efficiency of the quadtree approach. In addition, the proposed method does not alter the compression part of the traditional quadtree algorithm but its decompression (i.e. reconstruction) part only. This means that the proposed method could be utilized to improve the quality of the reconstructed image even if this image was compressed by the traditional quadtree approach.