Target Mask Reduction in Pulse-Doppler Radar Using Wavelet Domain Spatial Filtering
Aaron J. Wright, J. Patrick Donohoe · 2019
The shift-invariant properties of the redundant discrete wavelet transform (RDWT) produce spatial correlation between adjacent scales for significant features of a signal. In previous work, edges were detected in medical imaging based directly on the locations of high correlation. The technique presented in this work follows the same principle of spatial correlation between adjacent scales to identify interference in target detection of range-Doppler map applications. A filtering technique is presented to determine the location of interfering targets in constant false alarm rate (CFAR) detection and remove them as a means to reduce mutual target masking. In this paper, the filtering technique is applied to radar detection of 2-D range-Doppler maps produced by LFM pulse-Doppler radar. A CFAR detection algorithm using this filtering technique is presented and the performance is compared with other robust CFAR detectors. It is shown that the proposed CFAR detector utilizing the interference filtering technique performs better in suppressing target masking when compared to the other CFAR detectors.