Hyperspectral imaging holds enormous potential to improve thestate-of-the-art in aerial vehicle tracking with low spatial and temporalresolutions. Recently, adaptive multi-modal hyperspectral sensors, controlledby Dynamic Data Driven Applications Systems (DDDAS) methodology, have attractedgrowing interest due to their ability to record extended data quickly from theaerial platforms. In this study, we apply popular concepts from traditionalobject tracking - (1) Kernelized Correlation Filters (KCF) and (2) DeepConvolutional Neural Network (CNN) features - to the hyperspectral aerialtracking domain. Specifically, we propose the Deep Hyperspectral KernelizedCorrelation Filter based tracker (DeepHKCF) to efficiently track aerialvehicles using an adaptive multi-modal hyperspectral sensor. We address lowtemporal resolution by designing a single KCF-in-multiple Regions-of-Interest(ROIs) approach to cover a reasonable large area. To increase the speed of deepconvolutional features extraction from multiple ROIs, we design an effectiveROI mapping strategy. The proposed tracker also provides flexibility to coupleit to the more advanced correlation filter trackers. The DeepHKCF trackerperforms exceptionally with deep features set up in a synthetic hyperspectralvideo generated by the Digital Imaging and Remote Sensing Image Generation(DIRSIG) software. Additionally, we generate a large, synthetic, single-channeldataset using DIRSIG to perform vehicle classification in the Wide Area MotionImagery (WAMI) platform . This way, the high-fidelity of the DIRSIG software isproved and a large scale aerial vehicle classification dataset is released tosupport studies on vehicle detection and tracking in the WAMI platform.
展开▼
