Scalable low bit rate coding using an HC-RIOT coder

This paper discusses a low bit rate (LBR) wavelet based image coder that is designed to transmit over varying bandwidth network connections and is receivable by several different performance levels of decoders. It combines the discrete wavelet transform with a two stage ZTE/SPIHT zerotree entropy type coder to create a bitstream that is transmitted in base and enhancement layers with properties of progressive transmission, scalability, and perceptual optimization after a minimum bit rate is reached. A wavelet block chain (WBC) method is used in the initial ZTE coding of the low-low (LL) subband to separate and label homogenous regions of the image to the decoder without additional overhead information in the bitstream. This spatial information is then exploited to increase coding efficiency and perceptually improve the transmission of the image. This transmission is formatted such that an unequal channel protection can be applied. Lastly the new image coder has possibilities to assist in spatio-temporal segmentation method; and video object based coding.