In subband coding (SBC), an image is first filtered to create a set of images, each of which contains a limited range of spatial frequencies. These images are called the subbands. Since each subband has a reduced bandwidth compared to the original full-band image, they may be downsampled. This process of filtering and subsampling is termed the analysis stage. The subbands are then encoded using one or more coders. Different bit rates or even different coding techniques can be used for each subband, thus taking advantage of the properties of the subband and/or allowing for the coding errors to be distributed across the subbands in a visually optimal manner. Reconstruction is achieved by upsampling the decoded subbands, applying appropriate filters, and adding the reconstructed subbands together. This is termed the synthesis stage. Note that the formation of subbands does not create any compression in itself (since the same total number of samples is required to represent the subbands as is required for the original image). The motivation for this approach is that the subbands can be encoded more efficiently than the original image. A block diagram for a basic 1-D, two-band system is shown in Fig. 13.1, where the downsampling is by a factor of two. In this diagram, the analysis filters, h 1 (n) and h 2 (n) , are lowpass and highpass, respectively, and g 1 (n) and g 2 (n) are the corresponding synthesis filters.
The key elements in a SBC system are
- the analysis and synthesis filtering banks and
- the coding technique(s) applied to the subbands.
The analysis and synthesis filtering banks are first described for 1-D signals and then extended to 2-D signals.
Online access to SPIE eBooks is limited to subscribing institutions.