An adaptive synthesis filter bank for image decoding with fractional scalability

Transform image coding and more particularly the subclass of block based transformations are widely used to compress images. The JPEG standard for still images and MPEG codec specifications for video are very efficient implementations, but these algorithms perform image reconstruction without taking into account the quantization operations performed on the transform coefficients. In this paper, we propose an adaptive algorithm to tune the inverse transformation matrix as a function of the quantization level in order to minimize the reconstruction error. The developed algorithm provides quality scalability features and also integrates resizing operations into the inverse transformation process leading to a spatial scalability of fractional factors.