Registration of 3-D laser radar data and hyperspectral imagery for target detection

Authors:

  • Chan Kevin

Publish date: 2006-01-01

Report number: FOI-R--2101--SE

Pages: 54

Written in: English

Abstract

The purpose of this master´s thesis is to combine sensors for target detection and recognition, the sensors in this case being a hyperspectral camera and a scanning 3-D laser. Registration of the sensor data is essential in order to combine the sensors. The registration is performed by geometrical calibration of the sensors with repsect to each other. A problem with calibrating the sensors is that the scanning 3-D laser delivers 3-D point clouds and the hyperspectral camera returns 2-D images. Also, there is a wavelength problem. The hyperspectral camera has 240 different bands at its disposal, whereas the laser only uses one. This thesis is a part of a two-man project, where the other part is done as a master´s thesis by Christina Freyhult. Her part is to implement the anomaly detector and the visualization of the results. We have designed and implemented a method for automatic registration of 3D-laser data and hyperspectral imagery. We have thus shown that this kind of fusion is possible to realize, and we judge the potential to be tremendously high.