Runway Detection using Pushbroom Hyperspectral Camera

Abstract

Hyperspectral image analysis techniques use reflective solar radiation acquired by hyperspectral cameras to identify target present in the scene. A target can be any material surface of interest within an acquired image. The image acquisition mode of mostly available commercial hyperspectral camera is pushbroom. A pushbroom hyperspectral camera is a line scanning image sensor which collects ground image data parallel to the flight path at a specific point angle. The one dimension of acquired hyperspectral image represents the swath width while the other represents the spectral range. In this work, detection of selected materials present in the scene is performed. The spectral signature libraries of the target material and the background materials present in the scene are known. This prior spectral signatures knowledge is utilized in image bands selection to decrease processing and storage requirements while at the same time minimizing target and background visual blending. Moreover, post processing techniques making use of the expected context and size/orientation of the target are developed for a successful detection.