License Plate Extraction From Still Images

License plate extraction is an important stage in the vehicle license plate recognition for an automated transport system. This paper presents a novel and practical license plate extraction algorithm based on the edge statistics and the morphology. The proposed approach can be divided into four sections, which are vertical edge extraction, background curve and noise removing, edge statistical analysis and morphology-based license plate extraction. The proposed approach is designed to work in a wide range of acquisition conditions, including unrestricted scene environments, different camera-to-car distances, etc. Under the available database, the average accuracy of locating vehicle license plate is 84%.