A license plate super-resolution model reconstruction method and device

Abstract

The embodiment of the invention discloses a license plate super-resolution model reconstruction method and device, which are used for improving the license plate recognition rate. The reconstruction method comprises the following steps: step (1), acquiring a first image of a license plate, and performing low-resolution processing on the first image to obtain a second image; step (2), determining a first local image in the first image and a second local image in the second image; step (3), performing convolution processing on the second local image according to a convolution layer parameter of a first model to obtain a target local image; step (4), determining a pixel difference between the target local image and the first local image; step (5), judging whether the pixel difference satisfies a preset condition; and step (6), if yes, then adjusting the convolution layer parameter of the first model according to the pixel difference to obtain a target model, and if not, then adjusting the convolution layer parameter of the first model according to the pixel difference to obtain a second model, and taking a convolution layer parameter of the second model as the convolution layer parameter of the first model and repeatedly executing steps (2) to (6).

Description

technical field [0001] The invention relates to the technical field of image processing, in particular to a method and device for reconstructing a license plate super-resolution model. Background technique [0002] With the rapid development of modern transportation technology and economy, automobiles have become an important means of transportation in people's daily life, which inevitably has a higher requirement for traffic management. The automation of field management, etc., the research of license plate recognition technology has become a very important development direction. [0003] The license plate is the unique identity of a vehicle. Vehicle License Plate Recognition (VLPR) refers to the ability to detect vehicles on the monitored road and automatically extract vehicle license information (including Chinese characters, English letters, Arabic numerals and license plate colors) The processing technology is mainly based on digital image processing, pattern recogniti...

AI Technical Summary

Problems solved by technology

[0007] The embodiment of the present invention provides a method and device for reconstructing a license plate super-resolution model. The local area mapping relationship between high-resolution images and low-resolution images is trained through the first model, and after the target model is obtained, the low-resolution image can be reconstructed. The operation of deblurring and denoising solves the problem of license plate recognition error or missing car caused by unclear license plate area or the influence of light noise, and achieves the purpose of effectively improving the license plate recognition rate

Images