Study of the Algorithms for Image Matching in Intelligent Transportation System

Abstract

Vehicle identification is part of the basic topics of the Intelligent Transportation System (ITS). Due to the complex and changeable background of vehicle pictures and the strict requirements of vehicle detection and identification under a massive traffic flow, vehicle identification has become a very important part in intelligent transportation. It is helpful for us to carry out meaningful research on vehicle identification. The traditional SIFT algorithm, which creates 128-dimensional feature vectors and then matches the vectors, has so many shortcomings, such as the slow speed of the match and extraction of the features, and the mismatch of the parallel changes in the gray area. The application of the SIFT method combining with BBF algorithm and carried on research on the vehicle violation pictures that were taken by road cameras is discussed in this paper. The research optimized the algorithm and improved the accuracy and recognition speed.