Scalable Foveation Image Coding with Fuzzy System

This paper presents a method of the scalable foveation image coding with a fuzzy system. Our objective is to maximize foveated wavelet image quality index (FWQI). With the foveated visual sensitivity model, image regions are first prioritized. The image regions closed to the foveation points have higher priorities than those, which are far away from the foveation points. The discrete wavelet transform (DWT) is utilized to transform the considering image to the wavelet domain. The wavelet coefficients corresponding to different image regions are weighted differently based on their priorities. To achieve the objective, we use a fuzzy logic system and an iterative method to select the image coding parameters, which are a number of weighted wavelet coefficients (NWC) and a quantization parameter (QP). There are two inputs and one output in the proposed scheme. The inputs are a set of NWC and targeted bit per pixel used to encode an image. The output is an interval of the potential QP depended on a specific value of NWC. The QP providing the maximum FWQI given a value of NWC is selected. We iteratively search for the value of NWC and QP providing the highest FWQI. SPIHT codec is used to generate the scalable bitstream of the discrete wavelet coefficients. Our simulation results show that the proposed scheme provides better image quality compared to previous work in both objective and subjective quality.