Study On The Location And Recognition Of License Plate And Vehicle Logo Using Image Analysis

Vehicle recognition system (VRS) is an important component of the intelligent traffic management system. The VRS has been an active research area in digital image processing, pattern recognition and computer vision. The VRS is mainly focused on the license plate recognition. However, the integration of the license plate recognition (LPR) and vehicle logo recognition (VLR) can provide more accurate vehicle information, which will be very useful for vehicle crimes investigation.First, we review a large amount of vehicle recognition systems and methods used domestically and internationally. The technology of locating and recognizing license plates and vehicle logos were studied and compared. A VRS which integrates the license plate and vehicle logo recognition is proposed in this thesis. The license plate recognition module locates the license plate based on the HSV color model, and then license plate characters are segmented through the projection of character traits. The license plate recognition is then accomplished based on the template matching and fixed invariant moment. The rough location of the vehicle logo is estimated according to the relationship between license plate and its logo. The accurate location is then identified using the texture information of vehicle logo background and the features extracted using an improved histogram of oriented gradients. The gradients are calculated based on the fractional calculus. Finally, the clonal selection algorithm and K-nearest neighbor method are employed for the vehicle logo recognition. In order to show the effectiveness of this proposed algorithm,300 images of vehicles which consist of 17 different types of vehicle logos were tested in our experiments. The experimental results show that this system recognizes the license plate and vehicle logo rapidly and effectively. Furthermore, this method improves recognition accuracy comparing with the traditional methods.