License plate generation method based on generative adversarial network

Abstract

The invention relates to a license plate generation method based on a generative adversarial network. The method comprises the following steps: S1, obtaining a license plate character library and a background image used for generating a template license plate; s2, randomly selecting characters according to license plate composition, placing the characters into the background image by using OpenCV, synthesizing a template license plate, and performing image enhancement on the template license plate; s3, constructing a license plate generation network based on a coding-decoding structure; s4, constructing a training set based on the license plate image in the real scene and the template license plate image set; and S5, based on the training set, training a license plate generation network based on a coding-decoding structure by using an adversarial generation model to obtain a trained license plate generation network for generating a license plate image with a label. Based on an encoding-decoding network structure, the license plate image with the label can be generated according to the template license plate, and the problem that the label of the traditionally generated license plate image is random is solved.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a license plate generation method based on a generative confrontation network. Background technique [0002] In the automatic license plate recognition system, license plate recognition is a very critical part, and its performance affects the final recognition result. In the current license plate recognition method based on deep learning, the license plate dataset has a significant impact on the training of the recognition network. Through large-scale, wide-coverage training set training, the recognition network can have better generalization and stronger robustness. However, limited to the collection of license plate data sets, it faces various challenges such as collection difficulties, high labeling costs, and data privacy. It is difficult to obtain a large number of license plate datasets covering a wide range of scenarios. Therefore, it is urgently required to use image...

AI Technical Summary

Problems solved by technology

[0004] In view of this, the purpose of the present invention is to provide a license plate generation method based on generative confrontation network, based on the encoding-decoding network structure, so that the license plate image with labels can be generated according to the template license plate, which solves the problem of traditionally generated license plate images. Questions tagged random

Images