交通信号灯检测与识别算法研究

发布时间：2018-03-03 23:09

本文选题：交通信号灯　切入点：级联匹配　出处：《中南大学》2014年硕士论文　论文类型：学位论文

【摘要】：摘要：伴随着人类社会的发展,交通管理逐渐走向智能,而交通设备也将步入智能化的时代。无人车的上路及智能辅助驾驶的出现,无疑推动了交通系统智能化的进展。交通信号灯作为交通标识的重要部分,其智能的检测与识别有着不可估量的作用。本文围绕交通信号灯的检测与识别展开研究,并在此基础上,设计一套鲁棒性良好,同时能满足实时性要求的系统。本文的主要研究工作如下： 1.提出级联匹配算法,改进交通信号灯的检测过程。在不同的色彩空间,分开检测交通信号灯及其背板区域,并通过级联匹配,准确定位交通信号灯。此算法能够较好的解决交通信号灯装置被遮挡的问题。 2.改进了2DPCA算法,并结合Gabor小波变换,提取交通信号灯的特征。文中采用基于类内自适应加权平均的模块2DPCA算法,通过实验与普通的2DPCA算法比较,该算法具有更优越的性能,主要体现在类别鉴定能力更强,时间复杂度更低两方面。 3.将交通信号灯的检测及识别算法应用在实际问题中。本文构建了交通信号灯识别系统,该系统由输入模块、检测模块、识别模块、显示模块及通信模块构成。实验结果表明,对于识别交通信号灯,该系统具有较高的正确率,同时有较好的鲁棒性及实时性。图50个,表10个,参考文献58篇。
[Abstract]:Absrtact: with the development of human society, traffic management is gradually moving towards intelligence, and traffic equipment will also step into the era of intelligence. Traffic signal light, as an important part of traffic sign, has an inestimable effect on intelligent detection and recognition. This paper focuses on the detection and recognition of traffic signal light. On this basis, a system with good robustness and real-time performance is designed. The main research work of this paper is as follows:. 1. A cascade matching algorithm is proposed to improve the detection process of traffic signal lights. In different color spaces, traffic signal lights and their backplane areas are detected separately, and the traffic signals are matched by cascades. The algorithm can solve the problem that traffic lights are blocked. 2. The 2DPCA algorithm is improved, and the feature of traffic signal is extracted with Gabor wavelet transform. In this paper, 2DPCA algorithm based on intra-class adaptive weighted average is adopted. Compared with the ordinary 2DPCA algorithm, the algorithm has better performance. It is mainly reflected in two aspects: stronger ability of category identification and lower time complexity. 3. The detection and recognition algorithm of traffic signal lamp is applied to the practical problem. The system is composed of input module, detection module, identification module, display module and communication module. The experimental results show that the system is composed of three modules: input module, detection module, identification module, display module and communication module. For the recognition of traffic lights, the system has a high accuracy, good robustness and real-time performance. 50 figs, 10 tables, 58 refs.
【学位授予单位】：中南大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U491.51

【相似文献】