自然图像的边缘检测方法研究

发布时间：2018-06-14 19:40

本文选题：边缘检测 + 蚁群算法　；参考：《吉林大学》2017年硕士论文

【摘要】：图像一直以来都是人们认识和了解世界的主要来源,在以前人们主要对图像进行传递,而现在对图像进行处理,提取出其中的信息变得越来越重要了。边缘是图像最基本的特征,包含了图像的主要信息,边缘检测作为图像处理中的关键步骤,对后续图像处理工作,例如比较深入的特征描述、识别和匹配等都有着非常重要的作用。边缘检测的方法多种多样,本文主要针对自然图像,对其边缘检测方法进行研究。首先阐述了自然图像的边缘检测方法的研究背景和意义,对边缘检测的相关技术和算法的研究现状进行分析总结。并且对经典的边缘检测算法进行研究,分析了其基本原理和特征,并基于自然图像的检测结果对这些算法进行分析和比较。然后介绍了经典蚁群算法的原理及基于TSP问题的应用模型,并对基于蚁群算法的边缘检测算法的原理进行分析和阐述。接下来介绍了基于基本蚁群的边缘检测算法以及自适应分布式阈值策略,在此策略中,每一种蚂蚁都自身携带一个局部阈值并在图像中动态更新。然后,针对蚂蚁随机初始化会使每次实验得到不同的结果,且存在在非边缘区域存在较多蚂蚁影响效果的问题,本文提出了基于梯度的蚂蚁均匀分布策略,使得蚂蚁在初始化分布的时候,更多的分布在边缘区域。最后将实验结果进行对比和分析。最后针对自然图像中的城市道路图像,应用边缘检测方法,进行车道线的检测和识别。对输入图像进行预处理和特征区域提取后,运行多种边缘检测算法对相同的输入图像进行处理,并分析和比较实验结果,然后针对图像的干扰背景,通过分析车道线特性,提出了基于轮廓图的干扰背景去除的改进,使得在接下来的车道线检测中的准确性有了一定的提升。并在此基础上介绍了车道线物理性质的识别,对车道线的类型识别和司机的安全驾驶做出了贡献。
[Abstract]:Image has always been the main source for people to know and understand the world. In the past, people mainly transmitted the image, but now it is more and more important to process the image and extract the information from it. Edge is the most basic feature of the image, which contains the main information of the image. Edge detection is a key step in image processing. Recognition and matching play an important role. There are many methods for edge detection. In this paper, the edge detection methods of natural images are studied. Firstly, the research background and significance of natural image edge detection methods are introduced, and the research status of edge detection techniques and algorithms are analyzed and summarized. The classical edge detection algorithms are studied, and their basic principles and characteristics are analyzed, and these algorithms are analyzed and compared based on the detection results of natural images. Then the principle of classical ant colony algorithm and the application model based on tsp are introduced, and the principle of edge detection algorithm based on ant colony algorithm is analyzed and expounded. Then, the edge detection algorithm based on basic ant colony and the adaptive distributed threshold strategy are introduced. In this strategy, each ant carries a local threshold and updates dynamically in the image. Then, aiming at the problem that ant random initialization will result in different results in each experiment, and there are more ant effects in non-edge areas, a gradient based uniform ant distribution strategy is proposed in this paper. So that ants in the initial distribution, more distributed in the edge of the region. Finally, the experimental results are compared and analyzed. Finally, the edge detection method is used to detect and recognize the lane line for the urban road image in the natural image. After preprocessing and feature region extraction of the input image, the same input image is processed by a variety of edge detection algorithms, and the experimental results are analyzed and compared. Then, according to the interference background of the image, the characteristics of the lane line are analyzed. An improvement of interference background removal based on contour image is proposed to improve the accuracy of lane detection. On this basis, the physical property recognition of lane line is introduced, which contributes to the type identification of lane line and the safe driving of driver.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41

【相似文献】