A Fast and Noise-Tolerant Method for Positioning Centers of Spiraling and Circulating Vector Fields

摘要

Identification of centers of circulating and spiraling vector fields are important in many applications. Tropical cyclone tracking, rotating object identification, analysis of motion video and movement of fluids are but some examples. In this paper, we introduce a fast and noise tolerant method for finding centers of circulating and spiraling vector field pattern. The method can be implemented using integer operations only. It is 1.4 to 4.5 times faster than traditional methods, and the speedup can be further boosted up to 96.6 by the incorporation of search algorithms. We show the soundness of the algorithm using experiments on synthetic vector fields and demonstrate its practicality using application examples in the field of multimedia and weather forecasting.