Vision-based vehicle type (model) recognition is a hot topic in the domain of intelligent transportation systems. But it is difficult to recognize the exact type (model) of a vehicle due to the influence of some factors, for example, the view variations. In this paper, we present a robust system of recognition of the type (model) of vehicles from several frontal vehicle images. We use the height of the number plate as a reference to eliminate the zoom effect. From this reference, we extract several geometrical parameters (distance, surface, ratio) of decision, on bases of images taken in real conditions, were tested and analyzed. Employing this model, a distance error process allows measuring the similarity between an input instance and the data bases classes. The fusion of the three classifiers using the artificial neural networks (ANN) for each parameter allows showing the effectiveness of our process for the identification of the type of vehicle. We obtain taking a threshold of 90% a False Identification Ratio (FIR) of 1.6% and a False Rejection Ratio (FRR) of 3%. The network was tested and it was capable of classifying the type of vehicle of the taken database and a classification ratio of about 97% was obtained. And the minimized error percentages constitute an additional factor for the success of the verification system. According to these parameters, the rate of identification can reach 97% on a basis of a realistic data set of over 1000 images made up of 12 models and 9 classes of the type of vehicles. The results show that this approach achieves very high levels of both identification and verification performance.