%0 Journal Article
%T 基于单阶神经网络与特征融合方法的交通信号灯识别<br>Traffic Signal Recognition Based on Single-Order Neural Network and Feature Fusion Method
%A 王增荣
%J Journal of Image and Signal Processing
%P 79-85
%@ 2325-6745
%D 2020
%I Hans Publishing
%R 10.12677/JISP.2020.92010
%X <div style=\"text-align:justify;\">
	针对传统交通信号灯识别方法存在小目标检测准确率低、遮挡或反射情况下无法识别等问题，本文提出一种基于卷积神经网络的交通信号灯识别方法。采用单阶神经网络模型作为分类器，将单摄像头拍摄的彩色图像经过预处理后作为神经网络的输入，自动提取特征，通过特征融合的方法得出识别结果。通过实验验证了该方法的可行性。<br />
Aiming at the problems of traditional traffic signal recognition methods, such as small target detection with low accuracy and unrecognizability under occlusion or reflection, this paper proposes convolutional neural networks. A single-stage neural network model is used as a clas-sifier, and the color image captured by a single camera is used as the input of neural network after preprocessing. The features are automatically advanced in advance, and the recognition result is obtained by means of feature fusion. The feasibility of the method was verified by ex-periments.
</div>
%K 深度学习，卷积神经网络，交通信号灯识别，特征融合<br>Deep Learning
%K Convolutional Neural Network
%K Traffic Light Recognition
%K Feature Fusion
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=34478