New results on two and three dimensional directional filter banks

Recently, a family of multidimensional, non-separable filter banks known as directional filter banks were introduced. In subsequent works, these filter banks were shown to be effective for the detection and enhancement of linear features in noisy environments, resolution enhancement, and in subband image coding. In this paper, simplifying the design and implementation of the directional filter bank and also extending the directional filter bank constructs to the decomposition of three dimensional data such as video is focused upon. These new decompositions are generated by applying the two-dimensional decompositions utilized in the directional filter bank family along the different planes of the three dimensional signal. The three dimensional subband decompositions may be useful in motion detection and compensation for video sequence analysis and compression as well as linear and planar motion detection in noisy environments