Hybrid fast algorithm for S transforms

The S transform is useful in time-frequency analysis. In this paper, we propose a hybrid algorithm to implement it adaptively. Since the window size of the S transform varies with |f|, it is reasonable to use different algorithm for different frequency to implement it. In this paper, we use the sub IDFT algorithm in the low frequency region and the sectioned convolution algorithm in the high frequency region to implement the S transform. From simulation, our algorithm reduces 54% of the computation time and much improves the efficiency of the S transform.