Scalable Image Coding based on D-A DWT and Edgmentation

Scalable image coding based on a direction-adaptive discrete wavelet transform (D-A DWT) demonstrates outstanding compression performance in comparison to an equivalent system based on the classical DWT. In this paper we propose a scalable coding approach which combines D-A DWT with segmentation. Basically, the adaptation of the directional wavelet bases is controlled by a segmentation of the image in non-rectangular regions, obtained through an edgmentation procedure. The resulting region contours are simplified and approximated by polygons. The optimal lifting directions are then selected for each segment and at each resolution. Preliminary experimental results demonstrate that the proposed coding system encodes a segmentation of the input image and in the same time preserves significant performance advantages over an equivalent system employing the classical DWT.