Time-frequency representations of a signal can provide a useful means for obtaining parameter estimates for signals consisting of various chirps. We demonstrate the utility of including edge information extracted from these time-frequency representations when using a Hough transformation to perform this task. In particular, we show that using the edge information: (1) reduces the variance of the chirp parameter estimates in the case where the chirp signal has a single component; and (2) reduces the amount of spurious cross talk results when the signal has multiple chirp components. We further demonstrate a variation of our technique that detects the onset and duration of individual chirp components. We propose this technique as a fast preprocessing step for other algorithms such as maximum likelihood estimation which can provide very accurate parameter estimates. © 2004 Elsevier Inc. All rights reserved.
Bennett, N. N., & Saito, N. (2005). Using edge information in time-frequency representations for chirp parameter estimation. Applied and Computational Harmonic Analysis. Academic Press Inc. https://doi.org/10.1016/j.acha.2004.11.002