HYDETECTOR USING SPECTRAL LIBRARY INDEX FOR TAEGET DETECTION IN HYPERSPECTRAL IMAGE

초록

HSI(HyperSpectral Image) is a high dimensional data of which bandwidth ranges between visible region and ultra violet region. Recently, applications and spectral libraries for material classification, target detection, etc., have been developed. However, they do not provide indexing mechanism so that it takes lots of processing time to detect target. In this paper, we propose HyDetector for target detection in HSI which is a prototype system providing index structure for spectral library. A pyramid technique was used to construct index and k-nearest neighbor algorithm and integrated spectral matching filter was used for query processing. Experimental result shows that the execution time and the accuracy of the proposed system are 99.2% shorter and 61% better than those of traditional system without using index, respectively.