License plate pre-detection method based on improved bing algorithm

Abstract

The invention discloses a license plate pre-detection method based on an improved BING algorithm, which belongs to the field of license plate recognition. According to the specific shape of the license plate, the present invention improves the detection window of BING and enriches the detection features of BING, so that the recommendation frame given by BING can completely contain the entire license plate. Since the detection form of BING in the present invention has changed, it can It is not necessary to perform detection under images of different scales as in BING's original method, but only need to use the traditional pyramid method to achieve good detection results.

Description

technical field [0001] The invention belongs to the field of computer vision, in particular to the technical field of license plate recognition. Background technique [0002] With the rapid development of the economy, more and more Chinese people have their own private cars. Therefore, an intelligent traffic system is needed to help people manage traffic order. The license plate is the ID card of a motor vehicle. Once you find the license plate, you can find the only corresponding motor vehicle. Therefore, the license plate recognition system is an important component of the intelligent transportation system. License plate detection is the first and most important step in license plate recognition. A good license plate detection algorithm requires fast detection speed and high recall rate. Traditional license plate detection mainly has the following methods: [0003] 1) The license plate detection method based on edge detection and image morphology, the method based on ed...

AI Technical Summary

Problems solved by technology

[0004] 2) The color-based license plate location method, most of the license plates are blue or yellow, you can use the color area to detect the license plate, and use the color information to detect the license plate in many cases. It cannot be detected at all, and this algorithm is sensitive to light and cannot detect police car license plates (white)

Because the existing BING algorithm uses a sliding window of 8*8, it will result in the detected recommendation frame (the position identification of the detection target, usually a rectangular frame covering the detection target, whose coordinates pass through the upper left corner and right corner of the rectangle. The coordinates of the lower corner) does not just include a complete object, but usually only includes most of the target area. When the number of recommendation frames of a single image reaches a certain number, the detection rate is extremely low. For example, Dangdang single When there are 5,000 recommended boxes for a picture, the recall rate (recall rate) of the target object can reach 99.5%.

Images