复杂自然环境下的交通标志检测算法研究
发布时间：2019-01-09 14:18
【摘要】：近年来,越来越多的研究者开始关注智能交通系统(ITS),而交通标志检测是智能交通系统的重要环节,并且是交通标志识别的前提,具有重要的研究意义和应用价值。面对相对复杂的自然场景,交通标志检测目前尚没有成熟的实际应用,因此需要继续研究一种对于复杂场景下的交通标志普遍适用的检测方案。本文针对复杂环境下的交通标志检测算法进行研究,目的是提出一种对实际应用时的各种场景均适用的检测方案。本文的主要工作有如下几个方面:(1)对比分析几种基于颜色空间的交通标志分割方法,针对固定阈值颜色分割方法的缺陷提出一种基于RGB颜色空间的自适应阈值分割方法RGBAT(RGB Color Space Adaptive Threshold)。该方法能有效克服固定阈值分割方法易受光照、标志褪色、阴影等因素影响的缺陷。通过与几种基于颜色空间的交通标志分割方法进行对比实验分析验证了RGBAT方法的有效性。(2)依据交通标志检测应用对梯度直方图特征HOG与HSV量化直方图特征进行简化,在保证特征的分类能力的同时降低特征维数。(3)针对复杂环境下的交通标志图像提出一种新的普适的检测方案。对交通标志图像分为高亮度、中亮度与低亮度三类,每一类设计合适的检测方法;在交通标志精确检测步骤中对候选区域分为高亮度、中亮度、低亮度三类,针对不同类别分别训练出相应的SVM分类器:High_SVM、Medium_SVM、Low_SVM。对复杂的场景进行分类处理可以在一定程度提高检测方法的适应性。
[Abstract]:In recent years, more and more researchers have begun to pay attention to the intelligent transportation system (ITS), and traffic sign detection is an important part of the intelligent transportation system, and it is the premise of traffic sign recognition, which has important research significance and application value. In the face of relatively complex natural scene, traffic sign detection is not yet mature practical application, so it is necessary to continue to study a general detection scheme for traffic signs in complex scenarios. In this paper, the traffic sign detection algorithm in complex environment is studied. The purpose of this paper is to propose a detection scheme that is applicable to all kinds of scenarios in practical application. The main work of this paper is as follows: (1) several traffic sign segmentation methods based on color space are compared and analyzed. An adaptive threshold segmentation method based on RGB color space (RGBAT (RGB Color Space Adaptive Threshold).) is proposed to overcome the defects of the fixed threshold color segmentation method. This method can effectively overcome the defects of the fixed threshold segmentation method which are easily affected by illumination, flag fading, shadow and other factors. The effectiveness of RGBAT method is verified by comparing it with several traffic sign segmentation methods based on color space. (2) the gradient histogram feature HOG and HSV quantization histogram feature are simplified according to the traffic sign detection application. The feature dimension is reduced while the classification ability of the feature is guaranteed. (3) A new universal detection scheme for traffic sign images in complex environments is proposed. The traffic sign image can be divided into three categories: high brightness, medium brightness and low brightness. In the process of accurate detection of traffic signs, candidate regions are divided into three categories: high brightness, medium brightness and low brightness. The corresponding SVM classifier, High_SVM,Medium_SVM,Low_SVM., is trained for different categories. Classification of complex scenes can improve the adaptability of detection methods to some extent.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP391.41
[Abstract]:In recent years, more and more researchers have begun to pay attention to the intelligent transportation system (ITS), and traffic sign detection is an important part of the intelligent transportation system, and it is the premise of traffic sign recognition, which has important research significance and application value. In the face of relatively complex natural scene, traffic sign detection is not yet mature practical application, so it is necessary to continue to study a general detection scheme for traffic signs in complex scenarios. In this paper, the traffic sign detection algorithm in complex environment is studied. The purpose of this paper is to propose a detection scheme that is applicable to all kinds of scenarios in practical application. The main work of this paper is as follows: (1) several traffic sign segmentation methods based on color space are compared and analyzed. An adaptive threshold segmentation method based on RGB color space (RGBAT (RGB Color Space Adaptive Threshold).) is proposed to overcome the defects of the fixed threshold color segmentation method. This method can effectively overcome the defects of the fixed threshold segmentation method which are easily affected by illumination, flag fading, shadow and other factors. The effectiveness of RGBAT method is verified by comparing it with several traffic sign segmentation methods based on color space. (2) the gradient histogram feature HOG and HSV quantization histogram feature are simplified according to the traffic sign detection application. The feature dimension is reduced while the classification ability of the feature is guaranteed. (3) A new universal detection scheme for traffic sign images in complex environments is proposed. The traffic sign image can be divided into three categories: high brightness, medium brightness and low brightness. In the process of accurate detection of traffic signs, candidate regions are divided into three categories: high brightness, medium brightness and low brightness. The corresponding SVM classifier, High_SVM,Medium_SVM,Low_SVM., is trained for different categories. Classification of complex scenes can improve the adaptability of detection methods to some extent.
【学位授予单位】：河北师范大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：U495;TP391.41
【相似文献】
相关期刊论文 前10条
1 王歆s，
本文编号：2405739
本文链接：https://www.wllwen.com/kejilunwen/daoluqiaoliang/2405739.html
