License plate positioning method and device

Abstract

The invention provides a license plate positioning method and a license plate positioning device. The problem, which cannot be solved in the prior art, of template matching of relatively narrow characters, is effectively solved, and the license plate positioning precision is greatly improved. The license plate positioning method comprises the following steps: acquiring a license plate image, roughly positioning the license plate image, establishing a license plate template, and matching the license plate image which is subjected to rough positioning with the license plate template row by row, thereby obtaining the position of a license plate.

Description

Technical field [0001] The invention relates to the technical field of video image license plate positioning, in particular to a license plate positioning method and device. Background technique [0002] The license plate recognition system is one of the important components in the modern intelligent transportation system. It is a key technology to identify vehicle information. It plays an important role in public transportation, vehicle management, and even criminal investigation. The license plate recognition system mainly includes image capture and acquisition, license plate location, character segmentation, character recognition and other steps. Among them, the license plate location is an important basis for subsequent license plate recognition processing. [0003] The existing license plate location methods mainly include the location method based on vertical texture features, the morphology location method, the location method based on color features, the location method bas...

AI Technical Summary

Problems solved by technology

Considering that the vertical texture of the license plate image is relatively rich, many scholars use this method to locate the license plate, but in many cases the license plate is artificially disturbed, such as adding a license plate decoration frame, etc., causing the vertical texture near the license plate to be very rich, making this positioning method Received a lot of interference, the positioning accuracy is greatly reduced

[0008] Using the pattern recognition method positioning method for training and detection has a certain anti-interference ability, but the accuracy cannot be guaranteed very well. This is because the pattern recognition method positioning method will definitely involve the pyramid layer of the image, and the detected target target fusion, which is simply not enough for accurate positioning

[0009] Therefore, the scope of application of the above-mentioned license plate positioning methods is relatively narrow, and it is difficult to efficiently and accurately locate the license plate. In most cases, the above-mentioned method can only be used for rough positioning of the license plate

[0010] For this reason, a kind of method that license plate is carried out precise location appeared again in the prior art, this method needs to set up a license plate template of 409 (W) * 90 (H) pixel according to the size of standard license plate, as figure 1 and 2 As shown, the coordinates are used to express the position of each character of the standard license plate, and then the template matching is performed, but because some characters have a relatively narrow width, such as "1", most of its sides are the background, when performing template matching, the template The character area of ​​the narrower character does not match the position area on both sides of the narrower character, which will greatly reduce the accuracy of the license plate location, and then affect the subsequent license plate recognition process

Images