Highly scalable, low-complexity image coding using zeroblocks of wavelet coefficients

We propose a new highly scalable wavelet transform-based image coder, called S-SPECK, on the extension of a well-known zero-block image coder SPECK, by achieving not only distortion scalability, resolution scalability, and region of interest (ROI) retrievability, but also excellent compression performance with very low computational complexity. Though new features have been introduced into S-SPECK, our coder is quite competitive with SPECK on compression performance (peak signal-to-noise ratio) and computational complexity (encoding and decoding times) at various bit rates for standard test images. A novel quality layer formatting method is implemented in S-SPECK, which is much simpler and faster than PCRD used in JPEG2000. Extensive experiments have verified all our claims for S-SPECK.