Research On Multi-strategy Target Detection Methods On Complex Clutter Background

High Frequency Surface Wave Radar (HFSWR), using the high-frequencyelectromagnetic waves diffraction principle in the sea surface, is an effectivesolution to detection performance loss caused by the stealth technology forconventional microwave radar, achieves a large-scale, all-weather, remotemonitoring of ocean state and concerns in maritime surveillance for ship targetdetection. Due to a large number of various types of external noise and clutteroccupies part of HFSWR detection background, it shows a time-varying,multi-source, undulating complex characteristic. Only applying classical constantfalse alarm rate algorithm is easy to result in higher false alarm and poor detectionperformance, which makes HFSWR ship target detection become a difficultproblem.In order to effectively improve the detection performance for multi-targetcoexistence ship target, this paper research a adaptive multi-strategy target detectionmethod, which is based on background identification and classification. This methodincludes detection background analysis, background segmentation processing,background data index normalization and other parts, the main contents are asfollows:1, Study the HFSWR detection environment deeply, introduce its uneven andmulti-clutter features, and analyzes the generating principle and the physicalcharacteristics of the several main types of clutter, which can be the theoreticalgroundwork for the background segmentation based on type of clutter in laterarticle, and get the conclusions of the necessity of background perception andinformation extraction.2, Against multi-source, uneven complex features of the detection background,do segmentation processing for the detection background: use K-L divergenceprinciple to divide the detection background into homogeneous section andnon-homogeneous section; then divide the detection background into four regionsaccording to the type of the clutter, analyze each divided region based on statisticaltheory, it can be concluded that each partition data subject to Weibull distributionwith different parameter, through these studies, we achieve extracting usefulinformation and knowledge of the detection background, which can be use for targetdetection strategy design and parameter selection to complete multiple strategiesdetection system.3, In order to further improve the performance of detection strategies, researchHFSWR detection background pretreatment technology, using curvilinear regression theory to remove the big power points which influence statistics and backgroundregion segmentation. Introduce the background data exponential normalizationtechnique, which can normalize background data of the Weibull distribution intoexponential distribution data with same parameter, achieve uniform process todetection background, and can be an effective solution to the loss of detectionperformance caused by the differences in statistical for the detection background.This technique can enhance CFAR detection performance. Based on the above studya two-parameter CFAR detection method based on pre-segmentation is proposed,which do index normalization for each division area according to segmentationresults, then do the classical CFAR detection processing respectively, the detectionperformance is improved effectively.4, Summarize the above findings, we propose a multi-strategy target detectionstrategy on complex detection background for HFSWR, and detail algorithmprocesses of the strategy. Integrate the background segmentation technology, highpower points removal techniques, background data index normalization techniquetogether, then apply them into the detection strategy. After that, we analysisperformance of this strategy using the measured data added simulation targets, useseveral other widely used detection algorithm as a reference. The result of theexperiments shows that the ability to discovery weak target of this multi-strategymethod is stronger.Multi-strategy target detection on complex clutter background proposed in thispaper, specifically, effectively achieves the perception and application of thebackground real-time information, achieves a major breakthrough in the weak targetdetection, provides a reference for the research of radar intelligent processing in thefuture.