Naïve Bayesian classifier for human shape recognition

The aim of this study is to investigate the potential of Radon Transform and Regularized Principal Component Analysis as feature extraction for classification of pedestrian, non-pedestrian and vehicles. Several classification techniques are evaluated and verified based on accuracy, specificity and computational time. Initial findings showed that the best classification technique is Naïve Bayesian along with Gaussian as kernel with 100% accuracy and execution time of 0.016s respectively for human/vehicles classification while for pedestrian/non-pedestrian classifications are 97% respectively.