Efficient Elimination of False Detections Due to Sea Spikes in Coherent Radars
Anatolii A. Kononov
This paper introduces a method for efficiently eliminating false detections in coherent radar systems due to sea spikes. The proposed method employs scan-to-scan processing over a predefined number of successive antenna scans. Processing consists of matching estimated range-azimuth-velocity centroid associated with each radar plot extracted from a set of data detected within the current scan (initial plot) with centroids of radar plots generated in the previous scans. For each initial plot, the proposed method matches radar plots using a sequence of correlation windows generated in turn for each of the predefined previous scans. Each correlation window defines a range-azimuth region, with center and extent in range and azimuth adjusted from scan-to-scan. A group of matched plots is selected from all plots falling into a correlation window; these plots meet the velocity matching condition. Only the one radar plot, which minimizes the predefined overall matching criterion, is selected from the given group of matched plots for inclusion in the set of correlated plots associated with the initial plot. For each identified set of correlated plots, an overall correlation value is computed. If the correlation value exceeds a predefined threshold, the initial plot associated with that set of correlated plots is stored in memory for further processing and visualization. Otherwise, the initial plot is retained for plot rematching with modified velocity centroids. The modified velocities provide the detection of those targets of interest that may have been missed due to ambiguous radial velocity measurements. In contrast to known methods, the proposed method minimizes the correlation window area at a given probability of falling into the window for radar plots associated with a target corresponding to an initial plot. The proposed method efficiently eliminates the false detections while maintaining reliable detection performance for targets of interest; the detection performance is essentially improved compared to known methods. Additionally, the proposed method ensures reliable target detection when radial velocity measurements are ambiguous, a situation where known methods collapse.