| Directional analysis plays an important role in many areas and situations affecting our actual life such as oil exploration, medicine, remote sensing, and data analysis. There have been many proposed approaches for directional analysis in the past. The areas of applicability for this directional analysis covers almost all areas of 2-D and 3-D signal processing.; First, this thesis introduces a new directional filter bank based on the polyphase structure. Visual information in subbands is made available by use of a new structure of directional filter banks so that any spatial processing can be applied on visualizable subbands for more efficient and effective directional analysis. A technique to allow for linear phase filtering by using IIR filters in the directional filter bank is developed. This allowed the directional filter bank to have much more efficient computations than with FIR QMF filter prototypes that have similar filter characteristic. Also, a whole-sample delayed FIR filter prototype is examined and used in the directional filter bank for fast implementation in the case where perfect reconstruction is not a concern.; In addition to investigating the theory and design of the directional filter banks, applications of this new directional filter bank are also investigated. By virtue of visualizable subbands, more spatial processing could be applied. Two successful applications are introduced, which were an automatic target recognition system and a fingerprint enhancement system, based on the new directional filter banks. Also, by using separability of directional filter banks, a 3-D directional filter bank is proposed as a computationally attractive alternative in motion estimation and target tracking in complicate scenes. |
