Image processing method and device, equipment and medium

Abstract

The embodiment of the invention discloses an image processing method and device, equipment and a medium. The method comprises the steps of: obtaining a center point recognition result, an image type recognition result and coordinates of a reference point through recognition after a target image is input into a preset target model; when determining that the target image belongs to the license plate image based on the central point recognition result, calculating coordinates of key points corresponding to the image type of the target image according to the coordinates of the reference points; and positioning the target image by using the coordinates of the key points, and segmenting the target image of which the image type is a double-row license plate to obtain a single-row image, wherein when the image type is a single-row license plate, four key points are included, and when the image type is a double-row license plate, four key points in the previous row and four key points in the next row are included. Unification of line numbers and angles of single-layer license plates and double-layer license plates can be realized, and the problems of license plate segmentation errors and license plate confusion caused by license plate segmentation under the condition that the double-layer license plates are shielded can be solved.

Description

technical field [0001] The invention belongs to the field of intelligent transportation, and in particular relates to an image processing method, device, equipment and medium. Background technique [0002] In the process of smart city construction, license plate recognition is an important topic, and the accuracy of license plate recognition is particularly important. In order to improve the accuracy of license plate recognition, the input images should be as consistent as possible during the process of license plate recognition. However, at present, many license plate recognition input objects often include large-angle images, single and double license plates, complex license plate backgrounds, etc., which increases the difficulty of recognition. Without increasing the complexity of the license plate recognition model, the existing technology has optimized the license plate processing, but there are also some shortcomings, such as: single-line and double-line image type jud...

AI Technical Summary

Problems solved by technology

However, at present, many license plate recognition input objects often include large-angle images, single- and double-line license plates, complex license plate backgrounds, etc., which increase the difficulty of recognition

Without increasing the complexity of the license plate recognition model, the existing technology has optimized the license plate processing, but there are also some shortcomings, such as: single-line and double-line image type judgment and license plate key point detection serial, which will increase time overhead; In addition, after using the four key points of the license plate to correct the single-line and double-line license plate positioning, when the double-line license plate is segmented according to the national standard license plate, the segmentation effect is not good when the upper and lower sides of the license plate are blocked, which affects the subsequent license plate recognition effect.

Images