Multi-Resolution Image Segmentation with Border Smoothness for Scalable Object-based Wavelet Coding

This paper introduces a multiresolution image segmentation algorithm for scalable object-based wavelet coding applications. This algorithm is based on discrete wavelet transform and multiresolution Markov random field (MMRF) modelling. The major contribution of this work is to add spatial scalability and border smoothness in the segmentation algorithm usable for object-based wavelet coding algorithm. To optimize the segmentation/extraction of objects/regions of interest in all scales of the wavelet pyramid, with scalability constraint, a multiresolution analysis is incorporated into the objective function of MMRF segmentation algorithm. The proposed algorithm improves border smoothness in all regions, particularly in lower resolutions. In addition to scalability between objects/regions in different levels, the proposed algorithm outperforms the standard multiresolution segmentation algorithms, in both objective and subjective tests, in yielding an effective segmentation that supports scalable object-based wavelet coding.