A highly scalable speck image coder

Pearlman's SPECK image coding algorithm is a distortion scalable coder which achieves excellent compression performance with very low computational complexity. However, SPECK has no resolution and ROI (region of interest) salabilities. In this paper, we extend it to a novel highly scalable image coder named S-SPECK through an efficient strategy of grouping the wavelet coefficients according to their resolution levels and relationship with ROIs. The proposed new coder not only retains all the advantages of SPECK, but also implements resolution and ROI scalabilities, which, we believe, is a better alternative than JPEG2000 for some applications. Extensive experiments have been conducted to verify all our claims.