2-Step Scalar Deadzone Quantization for Bitplane Image Coding

Abstract

Abstract-Modern lossy image coding systems generate a quality progressive codestream that, truncated at increasing rates, produces an image with decreasing distortion. Quality progressivity is commonly provided by an embedded quantizer that employs uniform scalar deadzone quantization (USDQ) together with a bitplane coding strategy. This paper introduces a 2-step scalar deadzone quantization (2SDQ) scheme that achieves same coding performance as that of USDQ while reducing the coding passes and the emitted symbols of the bitplane coding engine. This serves to reduce the computational costs of the codec and/or to code high dynamic range images. The main insights behind 2SDQ are the use of two quantization step sizes that approximate wavelet coefficients with more or less precision depending on their density, and a rate-distortion optimization technique that adjusts the distortion decreases produced when coding 2SDQ indices. The integration of 2SDQ in current codecs is straightforward. The applicability and efficiency of 2SDQ is demonstrated within the framework of JPEG2000.Index Terms-2-step scalar deadzone quantization, general embedded quantization, bitplane image coding, JPEG2000.