Vehicle Classification Using Neural Network in Forward Scattering Radar

This thesis unveils the potential and utilization of Neural Network (NN) in radar applications for target classification. The radar system under test is a special of its kind and known as Forward Scattering Radar (FSR). FSR is a special type of bistatic radar which the transmitted energy is scattered by a target and the target is so close to the transmitter-receiver baseline. Recent works had shown that FSR can be effectively used for classification, but the result can be further improved by using advance classification method. To proof this, result from FSR experiment were used. The target used for this experiment is a ground vehicle which is represented by typical public road transport. New features from raw radar signal were determined and extracted manually prior to classification process using Neural Network (NN). Two types of features in the time and frequency domain signature were examined, namely time required for counting zero crossings, first main lobe width, second main lobe- width and the number of lobes. Multilayer perceptron (MLP) back-propagation neural network trained with back propagation algorithm was implemented and analyzed. In NN classifier, the unknown target is sent to the network trained by the known targets to attain the accurate output. Two tasks of classifications are analyzed. The first task is to recognize the exact type of vehicle, four vehicle types were selected: Vauxhall Astra, Renault Traffic, Vauxhall Combo and Honda Civic. The second task is to group vehicle into their categories: small, medium and large. The proposed NN provides high percentage of successful classification which is 90% and 98% of overall data was correctly classified in vehicle recognition and vehicle categorisation respectively. The result presented show that NN can be effectively employed in FSR system as a classification method.

Preview PDF FK_2009_21a.pdf Download (418kB)

Actions (login required)

View Item