Shiptrack Detection Algorithm Study

摘要

Shiptracks are known to be a relatively common phenomenon, often appearing inAVHRR channel 3 imagery as anomalous, curvilinear cloud lines. Despite their significance to remote ship surveillance studies, the formation mechanisms responsible for shiptrack production are still largely unknown and their specific characteristics still undefined. A shiptrack detection algorithm being developed at the Naval Postgraduate School seeks to objectively detect and locate shiptracks on AVHRR imagery. This algorithm is a major step in objectively defining specific shiptrack characteristics and automating the search for additional shiptrack examples. The purpose of this study was to outline the logic of the detection algorithm and present a subjective performance summary of its usefulness. After careful analysis of the algorithm output files from several full satellite passes, it was determined that the algorithm is capable of detecting up to 65% of the fresh shiptracks within a full pass AVHRR image with a false detection rate of only 1.31 tracks per million square kilometers. This performance is likely to improve further with continued work focused on designing adequate filters to categorically reject many of the recurring false detections.