基于卷积神经网络的交通路标检测v1.doc

摘 要 交通管理及交通安全问题正受到人们越来越多的关注。在此背景下，智能交通系统的概念应运而生。作为智能交通系统的一部分，交通标志检测系统在驾驶辅助、交通标志维护、自动驾驶等多方面具有重要作用。然而，真实交通场景复杂多变，光照条件、天气条件、局部遮挡、背景色相似干扰、阴影干扰等问题使交通标志检测系统的研究远未达到成熟卷积神经网络是将人工神经网络和深度学习技术结合而产生的一个新型人工神经网络方法，具有局部感知区域、层次结构化、特征抽取和分类过程结合的全局训练等特点，在图像检测领域获得了广泛的应用。 本文在对人工神经网络特别是卷积神经网络的基本概念和算法进行了总结和介绍的基础上，以经典的卷积神经网络模型为基础，将其应用到交通标志检测任务当中。检测的路标类型包括警告、禁止、指示等交通标志，其中分别含有不同的前景背景颜色及形状。首先，我们采用图像分割技术，将目标图片分割成许多小区域，然后将这些区域一次输入到已经训练好的神经网络中去，由此可以得知目标图像中是否含有路标。 关键字:交通标志检测图像分割卷积神经网络 ABSTRACT Traffic regulation and safety problem is getting more and more attentioUnder this background, the concept of Intelligent Transportation System(ITS) is presented. As a component of ITS, the traffic sign detection system plays an important role in driver assistance, traffic sign maintaining and automatically driving. However, in the complicated traffic scenes, the problems of different lighting condition, weather condition, partial occlusion, similar background color and shadow interfering make the research of traffic sign detection far from mature. Convolutional Neural Networks (CNN) is a technology that combines ANN and recent Deep Learning method, which is characterized by local receptive field, hierarchical structure, global learning for feature extraction and classification, has been applied to many image detection tasks. In this paper, we first give a thoroughly introduction on the basic concepts of ANN and CNN, based on the classic CNN, we apply it to the task of traffic sign detection. Detection types include warning signs, prohibition signs, direction signs, and each type has a different background color and foreground shapes .First, we use image segmentation to divide image into many small target areas, then we thrown these areas into the neural network which has been trained , so we can judge whether the detecting image has a traffic sign in it. By this method, we can detect traffic signs in complex environments without extracting features first. This method is robust, comprehensive and researcha