Research of Vehicle Recognition Method under SURF Feature and Bayesian Model

This paper investigates the vehicle recognition method based on the SURF algorithm and Bayesian classification model. This method is robust for it overcomes the problems of picture quality and shooting angle. Through the unsupervised Bayesian machine learning method, the vehicle type of unknown pictures can be recognized and fallen under the best matching type. The result shows that the vehicle rate is between 60% and 70%. When the number of feature points is between 180 and 210, we get the best result. The scale of training set is extremely important for perfecting the feature point information. Generally think, the training set is the bigger, the better.