面向智能交通的恶劣天气图像增强方法研究

发布时间：2018-04-19 06:09

本文选题：图像增强 + 雾霾散射模型　；参考：《沈阳工业大学》2017年硕士论文

【摘要】：智能交通系统提高道路的利用率、保证交通的安全性和改善驾驶员的舒适性,已成为城市未来交通运输的发展方向。车辆监控系统和交通监管系统作为智能交通系统的主要组成部分,要求系统具有较高的安全性和稳定性,但室外天气条件变换复杂,雾、霾、雨、雪、沙尘等天气对智能交通系统中监控摄像头的成像产生影响,限制智能交通监控系统的应用环境。智能交通监控系统在恶劣天气条件下拍摄的图像,直观上呈现亮度、对比度下降,客观参数上图像均方差和清晰度降低,严重影响系统在目标识别、交通监控等方面的使用。本文的目标是对恶劣天气下图像视频进行增强处理,降低恶劣天气对智能监控系统的成像影响。本文主要针对雾霾条件的图像进行增强处理。提出了多种雾霾天气图像增强方法,分析雾霾天气的物理特性和雾霾天气图像的客观参数特性,建立雾霾天气散射模型。提出的方法包括,以图像HSL颜色模式理论为基础的图像增强方法;基于光线传播独立性理论的图像增强方法;基于大气雾霾散射模型和亮度模型图像理论为基础的图像增强方法以及基于快速离散曲波变换的图像增强方法。对于雨天条件的数字图像,本文提出基于模糊C均值理论的图像增强方法,分析雨滴的光学特性,建立雨滴的光学模型,阐述图像增强方法步骤。本文提出的多种方法使用Matlab和Visual Studio 2013软件编程实现,在Visual Studio2013软件中使用OpenCV计算机视觉库。通过程序得到增强后的数字图像,对增强后的图像从主观和客观两个方面进行分析。对比处理前后图像的图像质量,评价图像质量采用图像均方差,清晰度和信息熵等客观参数。实验结果证实本文提出的四种雾霾图像增强方法,能明显提升雾霾天气下图像的图像质量;基于模糊C均值理论的雨天图像增强方法能有效去除图像中的雨滴,得到的增强图像保留了较多的图像信息,图像清晰程度明显提升,扩展了智能交通监控系统的应用场景。
[Abstract]:Intelligent Transportation system (its) has become the development direction of urban transportation in the future, which can improve the utilization ratio of roads, ensure the safety of traffic and improve the comfort of drivers.As the main component of intelligent transportation system, vehicle monitoring system and traffic supervision system require high safety and stability, but outdoor weather conditions are complicated, such as fog, haze, rain, snow, etc.The weather, such as dust and dust, has an impact on the imaging of surveillance cameras in intelligent transportation system, which limits the application environment of intelligent traffic monitoring system.The images taken by the intelligent traffic monitoring system under the bad weather condition show the brightness, the contrast drop, the image mean square deviation and the sharpness of the objective parameters decrease, which seriously affect the application of the system in target recognition, traffic monitoring and so on.The aim of this paper is to enhance the image and video in severe weather to reduce the imaging effect of bad weather on intelligent monitoring system.This paper focuses on the image enhancement of haze condition.Several enhancement methods of haze weather images are proposed. The physical characteristics of haze weather and the objective parameter characteristics of haze weather images are analyzed and the haze weather scattering model is established.The proposed methods include image enhancement method based on image HSL color pattern theory, image enhancement method based on ray propagation independence theory, image enhancement method based on image HSL color pattern theory, and image enhancement method based on ray propagation independence theory.Image enhancement method based on atmospheric haze scattering model and brightness model image theory and image enhancement method based on fast discrete Qu Bo transform.For the digital images of rainy days, this paper presents an image enhancement method based on fuzzy C-means theory, analyzes the optical properties of raindrops, establishes the optical model of raindrops, and expounds the steps of image enhancement.The methods proposed in this paper are implemented by Matlab and Visual Studio 2013 software, and OpenCV computer vision library is used in Visual Studio2013 software.The enhanced digital image is obtained by the program, and the enhanced image is analyzed from the subjective and objective aspects.Compared the image quality before and after processing, the image quality was evaluated with objective parameters such as image mean square error, clarity and information entropy.The experimental results show that the four haze image enhancement methods proposed in this paper can significantly improve the image quality in haze weather, and the raindrops can be effectively removed by the method based on fuzzy C-means theory.The enhanced images retain more image information, and the clarity of the images is improved obviously, and the application scene of the intelligent traffic monitoring system is expanded.
【学位授予单位】：沈阳工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP391.41

【相似文献】