Application of Optical Flow and Methods of Group Lee Representation in Problems of Detecting of Movements and Pattern Detection on the Images
The methods of image processing for problems of detection and the analysis of movements and detection of concrete objects moving in video surveillance systems and other applications are considered. The field of speeds received by means of optical streams allows to analyze complex situations with which does not consult classical difference method, for example, the analysis of movement in crowd, or allocation of the person on strong disturbance images. The field of speeds is broken on cluster which allocate separate elements in complex area. For the detection on the personnel of objects is proposed the method, based on the theory of the ideas of continuous groups, which makes it possible to carry out a Fourier analysis relative to special functions, in the best way fitted out for the specific transformation groups, for example, group of the motions of plane. The analogs of the theorem about the roll and of the properties of Fourier transform make it possible instead of the labor-consuming correlation method to manage with the calculation of the pair of integrals.
Zdzislaw Gosiewski and Zbigniew Kulesza
V.P. Makarychev "Application of Optical Flow and Methods of Group Lee Representation in Problems of Detecting of Movements and Pattern Detection on the Images", Solid State Phenomena, Vols. 147-149, pp. 582-587, 2009