Linear Filtering in Image Sequences
It is shown that linear and shift-invariant filters, acting on time and space simultaneously, can discriminate between image regions moving uniformly with different velocities, irrespective of contrast. A general scene ist formulated mathematically and its three-dimensional Fourier transform is calculated. In frequency space, the energy of a moving image region is concentrated on a plane. A filter, the frequency response of which is fitted to this plane, can detect the object and separate it from the differently moving background. Design techniques for recursive and nonrecursive filters are described, and simulation results demonstrate the ability of this method.
KeywordsSpatial Frequency Image Region Frequency Space Dynamic Scene Recursive Filter
Unable to display preview. Download preview PDF.
- R.J. Schalkoff, and E.S. McVey: ‘A Model and Tracking Algorithm for a Class of Video Targets’, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. PAMI-4, No. 1, Jan. 1982.Google Scholar
- R.R. Read, J.L. Shanks, and S. Treitel: ‘Two-Dimensional Recursive Filtering’, in: T.S. Huang (ed.): ‘Picture Processing and Digital Filtering’, 2nd ed., Topics in Applied Physics, Vol. 6 (Springer, Berlin, Heidelberg, New York 1979).Google Scholar