FFT and MCC algorithms comparison in the identification of tracers for atmospheric motion estimation

Abstract

Abstract:

            In the estimation of atmospheric motion vectors from a sequence of images (normally triplet) the suitable tracer selection is an important primary step. This suitability of the tracer is identified either by Maximum Cross Correlation Coefficient (MCCC) or Fast Fourier Transform (FFT) between the tracer and target window pixels. The FFT algorithm is much faster than MCC but useful for the cyclic (periodic) observations. FFT limits its usefulness in real time applications as clouds are not perfectly sinusoidal in nature. The present paper shows the comparative results of both algorithms for Kalpana -1 satellite imagery data. In spite of inherent limitations the FFT based technique show significant improvement in computing time which is very useful in operational scenario of atmospheric motion vectors from the geosynchronous satellite data.

Key words: FFT, MCCC, Kalpana -1