LIDAR SIGNAL PROCESSING TECHNIQUES : CLUTTER SUPPRESSION, CLUSTERING AND TRACKING

Light detection and ranging (Lidar) is an object detection system, which uses the scattering of light, normally in the form of laser, to determine the direction, range, and other properties of a distant target. Due to the short wavelength of the laser beam, compared to traditional radio based ranging systems (radar), atmospheric phenomenon such as precipitation and fog present a major problem due to the water droplets reflectiveness at these wavelengths. Another problem encountered is that the output from the Lidar is presented as sets of two-dimensional data points, quantifying this information into discrete set of targets present a challenge. The third problem encountered in this field is that of data association, how to associate the target observations in space and time. The data output from the Lidar was analyzed and filters where developed to counter these problems. The problem of clutter suppression was solved using heuristic methods with knowledge of the clutter distribution. The data quantization problem mandated a modified solution to an existing algorithm, while the data association (tracking) problem was solved using an implementation of the Multiple Hypotheses Tracker (MHT) algorithm. It was shown to be possible to achieve a substantial improvement to the Lidars performance by the use of smart filters. In conclusion, this thesis contributes with an interesting approach to the problems encountered in Lidar signal processing