Locating car license plates using neural networks

The authors present a method for locating car license plates using neural networks. Neural networks are used as filters for analysing small windows of an image and deciding whether each window contains a license plate. A post-processor combines these filtered images and gives the final location of the license plates. The method offers robustness when dealing with noisy images. Tests with car images travelling on the road and at the entrance of a car park showed extraction rates of 99 and 97.5%, respectively. These results suggest that the proposed method works well with real-world situations.