Range-Doppler processing with the cone kernel time-frequency representation

Range-Doppler processing is the primary tool in active sonar and radar imaging. The matched-filter is the optimal detector in sonar and radar, but has drawbacks in time and frequency resolution which may hinder its performance as a classifier. The authors discuss a complementary processing methodology, the cone kernel time-frequency representation, that, although slightly suboptimal in a detection sense, yields simultaneously good resolution in time and frequency (range and Doppler). This processing may provide improved performance for target classification, especially in low signal-to-noise (SNR) environments, where the cone kernel processing outperforms the matched filter in Doppler estimation.<>