A method for clearing license plate images based on convolutional neural network

Abstract

The invention discloses a license plate image clearing method based on a convolutional neural network, which can clear a license plate image affected by motion blur. The method includes the following steps: establishing a lightweight convolutional neural network model, Set the structure and training parameters of the network; make the training set of convolutional neural network; train the convolutional neural network to obtain the weight of the network; cleared. The present invention is applied to criminal investigation and OCR recognition, etc., and for a road monitoring device, because the moving speed and appearance position of the vehicle are different, the blurring degree and blurring angle of the license plate image change within a certain range, and the traditional motion blurring method cannot be effective Processing, the method of convolutional neural network can clear the motion blur within a certain range of the license plate image, and has a faster processing speed.

Description

technical field [0001] The invention belongs to the field of image restoration and machine learning, and in particular relates to a method for clearing license plate images based on a convolutional neural network. Background technique [0002] In the process of identifying vehicles with license plate numbers, the imaging of road monitoring equipment is affected by various factors such as movement, light, and temperature, and the captured license plate numbers are prone to various types of blurring. The sharpening of fuzzy images has important applications in criminal investigation and OCR. In order to solve the problem of clearing the license plate image affected by motion blur, there are some deblurring methods based on probability statistics and digital image processing, such as the image clearing method based on the L0 norm prior, the blindness method based on the Lucy‐Richardson algorithm. Deconvolution methods, etc., these methods require manual adjustment of parameter...

AI Technical Summary

Problems solved by technology

For the vehicle images captured by the road monitoring equipment, due to the uncertain movement speed of the vehicle and the position of the vehicle in the field of view of the monitoring equipment, the captured license plate images will be subject to different degrees of motion blur, and the angle of the blur kernel is within a certain range. It is difficult for traditional image sharpening methods to effectively deal with this kind of image blurring.

Images