License Plate Extraction in Low Resolution Video

Extensive study has been conducted in the detection of license plate for the applications in intelligent transportation system (ITS). However, these results are all based on images acquired at a resolution of 640 times 480. In this paper, a new method is proposed to extract license plate from the surveillance video which is shot at lower resolution (320 times 240) as well as degraded by video compression. Morphological operations of bottom-hat and morphology gradient are utilized to detect the LP candidates, and effective schemes are applied to select the correct one. The average rates of correct extraction and false alarms are 96.62% and 1.77%, respectively, based on the experiments using more than four hours of video. The experimental results demonstrate the effectiveness and robustness of the proposed method