A method based on the convolution energy for testing vehicle license plate area

To test the vehicle license plate area is a main task based on video surveillance in ITS, especially for high-resolution video streams, now many kinds of methods are invalid for their complexity and slow speed. In this paper a new method was proposed to accomplish this task based on convolution energy. Firstly set an approximate detection area and generate brightness curve from the difference image projection. Through this curve the approximate position region of the vehicles can be located meanwhile the background model has been updated. Secondly take the advantages of the characters that the license plate owned complicated texture and sharp-cut contrast exterior to get the convolution energy figure, according to the vehicle license plate area possessed the maximum convolution energy to fulfill the testing of it. By this way on the one hand it can resolve the time-consuming problems when using background modeling to processing the high-resolution image, on the other hand it can also avoid to calculating the whole image convolution energy to achieve the purpose of real-time. Experiments show that this method has higher accuracy and robustness.