License plate detection method and device in complex scenes

Abstract

The invention discloses a license plate detection method in a complex scene, which includes: performing rough positioning on an input image, the rough positioning includes horizontal filtering and binarization processing, and using a trained fine positioning classifier to classify areas with rough positioning The license plate area detection is performed on the image of the license plate area, which also includes: the image of the license plate area is detected through the false detection classifier obtained by branch training to exclude the image of the falsely detected license plate area, and obtain the final image containing the license plate area. The invention also includes a license plate detection device. Compared with the existing methods, the license plate detection method and device adopted by the present invention have the outstanding advantages of low false detection rate and faster speed than the existing methods on the basis of ensuring the correct detection rate of the license plate.

Description

technical field [0001] The invention relates to the field of image processing, in particular to a license plate detection method and device in complex scenes. Background technique [0002] License plate detection technology is the core basic module in intelligent transportation systems, providing key information for many applications, such as vehicle capture, license plate location and recognition, vehicle violation capture, etc. [0003] In the actual traffic scene, the license plate is rich and diverse: there are blue cards and yellow cards; under different lighting conditions, there will be overexposure and low contrast; some vehicles will be blocked and dirty; license plates will be tilted at different shooting angles and other complex situations; and Some road and car body advertisements have rich textures, which are easy to try and misdetect. How to improve the license plate detection index and performance in various complex scenarios is a challenging problem. [000...

AI Technical Summary

Problems solved by technology

The disadvantage of this method is that there is no rough positioning, and it takes a lot of time to scan the entire vehicle body area for license plate detection, and there is no feature design for the license plate. Therefore, for license plates with rich body textures, there is also a high false detection rate.

Images