RECOGNITION OF BLACKLISTED VEHICLE BASED ON SIFT FEATURE

With the rapid development of technology, the intelligent traffic system is an important tool for traffic police in combating crime involving vehicles, In recent years, the recognition of blacklisted vehicle models is becoming more and more essential for crime tracking. However, it can be proved that the classical method of recognition, SIFT, has low accuracy when there are plenty of feature points with noise. In addition, current work in this field only focus on a few fuzzy classifications with not much practical value and testing in only one assessment method that doesn't have comprehensive representativeness. This paper proposes a novel method by using template library based on SIFT feature in order to solve these problems. The method pick up some commonly used vehicle models and test performance of algorithm based on Confusion matrix, Confirmed by the experiments, the proposed method has higher accuracy and only has a tiny delay.