Low bit rate image coding with shift orthogonal filter banks

Two critical issues that effect image quality as well as system design in wavelet image coders are the choice of wavelet bases and the quantization/coding strategy. In recent years orthogonal and biorthogonal wavelet bases have been successfully used for image coding. In this paper a class of spline based shift-orthogonal wavelet packets are used to achieve high quality compression at low bit rates. These shift-orthogonal wavelet bases allow the design of shorter wavelet synthesis filters while preserving the orthogonality within the independent wavelet channels. These features can be used for efficient quantization of individual channels that may result in reduction of reconstruction artifacts such as ringing effects at lower bit rates. Moreover by dropping the high frequency channels as requested by our multi-level EZW coding scheme, it is possible to create a highly scalable encoded bit stream