COMPARISON AND EVALUATION OF SHIP TARGET DETECTION ALGORITHMS WITH SAR IMAGES

摘要

Nowadays many ship target detection algorithms havernbeen developed based on SAR images. The typicalrnalgorithms include constant false alarm rate algorithm,rnwavelet transform algorithm and cross correlationrnalgorithm. But there are no standard test methods torncompare them. So in this paper, the three detectionrnalgorithms are used to extract ship signatures on arncommon set of ENVISAT ASAR images. Then thosernresults are compared with in situ measurement data forrnanalyzing the application scope and detection precisionsrnof each ship target detection algorithm. The comparisonrnshow that choice of an appropriate threshold probablyrnconstitutes one of the major drawbacks of constant falsernalarm rate algorithm, and constant false alarm raternalgorithm is not suitable in coarse sea state; On therncontrary, the wavelet transform algorithm and crossrncorrelation algorithm have great robustness and canrnextract several weak signatures which are easy to berndisturbed by background. But they can miss somerndetails information of vessel structure. Those details arernvery important for future vessel classification.