The number of vehicles in the urban areas is rising at high pace. The critical issues are arising with the rise in the number of vehicles for the traffic analysis. The analysis of the vehicle running across the roads is usually done for the density analysis, traffic shaping and many other similar applications. The vehicle detection in the rushed areas produces the real challenge of independent component selection and classification, which requires the precise object detector with deep analytical ability based classification algorithm. In this paper, the unique method with probabilistic neural network (PNN) classification model along with the non-negative matrix factorization for the purpose of vehicular object localization and classification in the urban imagery. The proposed model is expected to solve the problems associated with the accuracy, precision and recall.