The wavelet transform has become the most interesting new algorithm for still image compression. Yet there are many parameters within a wavelet analysis and synthesis which govern the quality of a decoded image. In this paper, we discuss different image boundary policies and their implications for the decoded image. A pool of gray-scale images has been wavelet-transformed at different settings of the wavelet filter bank and quantization threshold and with three possible boundary policies. Our empirical evaluation is based on three benchmarks: a first judgment regards the perceived quality of the decoded image. The compression rate is a second crucial factor. Finally, the best parameter settings with regard to these two factors is weighted with the cost of implementation. Contrary to the JPEG2000 standard, where mirror padding is implemented, our investigation proposes circular convolution as the boundary treatment.
Dieser Eintrag ist Teil der Universitätsbibliographie.
Search Authors in