Object Detection and Classication in Aerial Hyperspectral Imagery using a Multivariate Hit-or-Miss Transform

摘要

High resolution aerial and satellite borne hyperspectral imagery provides a wealth of information about an imagedscene allowing for many earth observation applications to be investigated. Such applications include geologicalexploration, soil characterisation, land usage, change monitoring as well as military applications such as anomalyand target detection. While this sheer volume of data provides an invaluable resource, with it comes the curse ofdimensionality and the necessity for smart processing techniques as analysing this large quantity of data can be alengthy and problematic task. In order to aid this analysis dimensionality reduction techniques can be employedto simplify the task by reducing the volume of data and describing it (or most of it) in an alternate way. Thiswork aims to apply this notion of dimensionality reduction based hyperspectral analysis to target detection usinga multivariate Percentage Occupancy Hit or Miss Transform that detects objects based on their size shape andspectral properties. We also investigate the effects of noise and distortion and how incorporating these factors inthe design of necessary structuring elements allows for a more accurate representation of the desired targets andtherefore a more accurate detection. We also compare our method with various other common Target Detectionand Anomaly Detection techniques.