Rate-distortion based image segmentation using recursive merging

In this paper, a rate-distortion based image segmentation algorithm for seg mentation-based coding is presented using a recursive merging with region a djacency graph (RAG). In the method, the dissimilarity between a pair of ad jacent regions is represented as a Lagrangian cost function considered in a rate-distortion sense. Lagrange multiplier is estimated in each merging st ep, a pair of adjacent regions whose cost is minimal is searched, and then the pair of regions are merged into a new region. The merging step is recur sively performed until some termination criterion is reached. The proposed method thus is suitable for region- or segmentation-based coding. Experimen tal results for 256 x 256 Lena show that segmentation-based coding using th e proposed method yields PSNR improvement of about 3.5 similar to4.5 dB, 1. 8 similar to2.0 dB, and 1.2 similar to1.5 dB over mean-difference based met hod, distortion-based method, and JPEG, respectively.