A Vehicle License Plate Locating Approach Based On Synthetic Characteristics

The automatic recognition of vehicle license plate is an important step in Intelligent Transportation System. The location of vehicle license plate is the most pivotal technology in recognition system, which plays an important role in the performance of whole system. This paper focuses on the key problem, and attempts to find a more effective vehicle license plate locating algorithm. Based on a through analysis of the present algorithms, a synthetic approach for vehicle license plate location based on color segmentation and texture analysis is proposed. This proposed approach combines the color feature, texture feature and geometric feature of vehicle license plate, which employs HSV color model and focuses on matching background color and character color of vehicle license plate to segment the car image with a simple and effective algorithm. Aimed at the segmented image, the background and character of vehicle license plate have the characteristic of consecutive grey beeps in horizontal direction. We can use line scanning to determine the top and the bottom boundary of vehicle license plate. Aimed at the car image with the top and the bottom boundary segmented, the vehicle license plate have projection characteristic in vertical direction. We can use differentiate sciagraphy to determine the left and the right boundary of vehicle license plate. Finally, by calculating the HVR to judge whether the area is the true vehicle license plate. The algorithm employs synthetic characteristics to locate license plate, which overcomes the defects and limitations in single characteristic locating license plate, and is slightly confined to the size or the position of the license plate and the illumination condition of the car image. Experiment results show that the locating approach is effective, high-precision and adaptable.