复杂场景下基于MSER-Sobel纵向算子的车牌实时检测方法  

Real Time License Plate Detection Method Based On Mser-Sobel Longitudinal Operator In Complex Scenes

作　　者：黄梅娟[1] HUANG Mei-juan(School of Information Engineering,Anhui Business And Technology College,Hefei,Anhui,231131)

机构地区：[1]安徽工商职业学院信息工程学院,安徽合肥231131

出　　处：《贵阳学院学报(自然科学版)》2024年第4期32-37,共6页Journal of Guiyang University:Natural Sciences

基　　金：安徽省职业与成人教育学会教育教学研究规划课题(AZCJ2023113);安徽省质量工程项目(2022zmgj005);安徽省自然科学基金(KJ2021A1511)

摘　　要：为改善复杂场景下实时车牌检测性能,提出了结合MSER和Sobel纵向算子的车牌检测方案。在预处理阶段,提出改进的对比度限制自适应直方图均衡化(CLAHE)技术,利用纹理和灰度特征检测车牌区域,解决了对照明和噪声的敏感问题,显著降低无用特征数量。为解决尺度不变性利用Sobel的提取结果帮助MSER得到复杂背景下的候选车牌区域。此外,在车牌候选区域内使用包围框调整技术,通过候选区域的累积强度直方图完成精细化调整,实现车牌区域的精准定位。实验结果表明,所提方法能以极快的速度准确完成车牌检测,在难度较大的AOLP-RP子集和CCPD-Challenge子集上分别取得了99.05%和92.9%的车牌检测准确度,显著优于其他优秀方法。To improve the performance of real-time license plate detection in complex scenes,a license plate detection scheme combining MSER and Sobel longitudinal operators is proposed.In the preprocessing stage,an improved contrast constrained adaptive histogram equalization(CLAHE)technology is proposed,which uses texture and gray features to detect the license plate area,solves the problem of sensitivity to lighting and noise,and significantly reduces the number of useless features.In order to solve the scale invariance,Sobel's extraction results are used to help MSER get the candidate license plate region under complex background.In addition,the bounding box adjustment technology is used in the candidate area of the license plate,and the refined adjustment is completed through the cumulative intensity histogram of the candidate area,so as to achieve the accurate location of the license plate area.The experimental results show that the proposed method can accurately complete the license plate detection at a very fast speed,and has achieved 99.05%and 92.9%license plate detection accuracy on the difficult AOLP-RP subset and CCPD-Challenge subset,respectively,which is significantly better than other excellent methods.

关 键 词：实时车牌检测 对比度限制 Sobel纵向算子 直方图均衡化 包围框调整 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]