Horn Extraction in Noisy Environments by Empirical Mode Decomposition

In this study, we propose a car horn extraction method in a noisy environment. In the proposed method, we use the empirical mode decomposition (EMD) and Hilbert transform for converting original signals to analytic signals. We can synchronize phases of reference and input signals by converting to analytic signals. The degree of the similarity of reference and input signals are defined as the inner product of them. Therefore, we can judge whether input signal is alert signal by the value of their inner product. Finally, in order to show the effectiveness of the proposed method, we demonstrated some simulation. Then, it was confirmed that the proposed method works very well.