关键词: 特征匹配, 尺度不变特征变换, 车标识别, BP神经网络

Abstract: Aiming at the problems that the matching threshold was difficult and the recognition speed was slow in the process of vehiclelogo recognition, we proposed a vehicle[KG-*4]\|logo recognition algorithm based on feature matching of scale invariant feature transformation (SIFT). The SIFT operator was used to extract the invariant features of the image, such as viewing angle, translation, radiation, brightness and rotation, and the BP neural network algorithm was used to autonomously select vehiclelogo image features for classification, matching and recognition. The results of simulation experiment show that the mean values of recognition rate for simple vehiclelogos and complex vehiclelogos are all more than 90%, the algorithm  has faster recognition speed and higher recognition rate, which can meet the needs of practical application.

Key words: vehiclelogo recognition； scale invariant feature transformation (SIFT)； feature matching； BP neural network,