Abstract: In view of the features of car types, this paper proposes a recognition method based on the BP neural network: first, pick up feature vectors of car types; then, we design the classifier using BP, and train it effectively; finally, this BP classifier can recognize the test samples. This paper mainly trains networks using the SCG algorithm. This algorithm can make the network converge quickly, thus solving the limitation of slow convergence of the common algorithm.

Key words: BP neural network, classifier of can types, BP(Back Propagation) algorithm