针对雾霾天气下交通信号灯定位准确率较低、图像增强时出现图像亮度不均匀的问题,该文提出一种基于改进的带色彩恢复的多尺度视网膜增强(Multi-Scale Retinex with Color Restoration,MSRCR)的雾霾天气下信号灯识别算法。首先利用改进的...针对雾霾天气下交通信号灯定位准确率较低、图像增强时出现图像亮度不均匀的问题,该文提出一种基于改进的带色彩恢复的多尺度视网膜增强(Multi-Scale Retinex with Color Restoration,MSRCR)的雾霾天气下信号灯识别算法。首先利用改进的MSRCR算法对有雾图像进行预处理,校正图像亮度并丰富图像细节;再利用最大稳定极值区域(Maximally Stable Extremal Regions,MSER)算法以及信号灯的背板信息确定信号灯的位置;最后将定位区域转换至HSV空间进行信号灯识别。结果表明,该方法能够在雾霾条件下有效地定位及识别交通信号灯。展开更多
针对目标检测算法在交通标志检测中存在的不足,文中提出了一种融合感受野增强模块和注意力机制的交通标志检测算法。该算法在YOLOv5(You Only Look Once version 5)算法的基础上改进,选用感受野模块(Receptive Field Block,RFB)替换原...针对目标检测算法在交通标志检测中存在的不足,文中提出了一种融合感受野增强模块和注意力机制的交通标志检测算法。该算法在YOLOv5(You Only Look Once version 5)算法的基础上改进,选用感受野模块(Receptive Field Block,RFB)替换原骨干网络中的空间金字塔池化(Spatial Pyramid Pooling,SPP)模块,在特征融合网络中嵌入高效通道注意模块(Efficient Channel Attention Module,ECAM)和卷积块注意模块(Convolutional Block Attention Module,CBAM),选用矩阵非极大值抑制(Matrix Non-Maximum Suppression,Matrix NMS)筛选候选框以提升算法的检测精度和检测速度。实验结果表明,在模型参数量与原网络相比未变化的前提下,该算法的均值平均精度达到了82.31%,与原算法相比提升了8.59%,检测速度达到了51.89 frame·s^(-1),且该算法在各个测试场景中未出现错检漏检现象,证明其泛化能力优于原算法,可以实时检测交通标志。展开更多
为提高智能网联(connected and automated,CA)卡车、小车及人工驾驶卡车、小车的混合流道路通行能力,提出基于排强度和渗透率的CA车辆单独编队和合作编队策略.分别设计两种策略下混合流车辆跟驰模式,推导出基于改进Markov模型,涵盖CA车...为提高智能网联(connected and automated,CA)卡车、小车及人工驾驶卡车、小车的混合流道路通行能力,提出基于排强度和渗透率的CA车辆单独编队和合作编队策略.分别设计两种策略下混合流车辆跟驰模式,推导出基于改进Markov模型,涵盖CA车辆渗透率和排强度的车辆状态转移概率;分析两种策略下CA车辆队列分布,建立各策略下的混合流道路容量模型,并通过理论证明和仿真实验予以验证.结果表明,与不编队策略相比,两种策略下道路容量分别提高1.23%~49.62%和1.47%~60.34%,合作编队策略与单独编队策略相比能将道路容量再提高11%;当CA车辆渗透率大于50%和排强度大于0时,编队策略对道路容量的提升效果更显著,容量能提高13.27%~60.34%;单独编队策略下CA小车和CA卡车最大队列规模分别为8辆和6辆,合作编队下CA车辆最大队列规模为8辆.展开更多
The modeling of network traffic is important for the design and application of networks, but little is known as to the characteristics of distribution of packets in network traffic. In this letter the distribution of ...The modeling of network traffic is important for the design and application of networks, but little is known as to the characteristics of distribution of packets in network traffic. In this letter the distribution of packets in network traffic is explored.展开更多
To measure the length of traffic queue, a vehicle motion model at intersections was built, and based on it the effective traffic queue was def＇med. Color images segmentation and frame differencing technique were used...To measure the length of traffic queue, a vehicle motion model at intersections was built, and based on it the effective traffic queue was def＇med. Color images segmentation and frame differencing technique were used to detect the foreground and the moving vehicles by detecting regions of the images, and then measure the length of effective traffic queue. By processing the image sequence acquired at certain intersection, the results prove that it is able to work out the traffic queue effectively by using the two techniques.展开更多
With the rapid development of highway transportation in China,intelligent transportation has become an important part of the traffic structure,and wireless sensor networks are also widely used in intelligent transport...With the rapid development of highway transportation in China,intelligent transportation has become an important part of the traffic structure,and wireless sensor networks are also widely used in intelligent transportation.However,in the wireless traffic sensor network,there is a certain error in the positioning of the anchor blind nodes.In the process of tracking the feedback information,the results of determining the position are very different.Based on the maximum degree of tension,the road traffic wireless Research and analysis of blind node location in sensor networks,and propose solutions and measures to reduce monitoring results.展开更多
文摘针对雾霾天气下交通信号灯定位准确率较低、图像增强时出现图像亮度不均匀的问题,该文提出一种基于改进的带色彩恢复的多尺度视网膜增强(Multi-Scale Retinex with Color Restoration,MSRCR)的雾霾天气下信号灯识别算法。首先利用改进的MSRCR算法对有雾图像进行预处理,校正图像亮度并丰富图像细节;再利用最大稳定极值区域(Maximally Stable Extremal Regions,MSER)算法以及信号灯的背板信息确定信号灯的位置;最后将定位区域转换至HSV空间进行信号灯识别。结果表明,该方法能够在雾霾条件下有效地定位及识别交通信号灯。
文摘针对目标检测算法在交通标志检测中存在的不足,文中提出了一种融合感受野增强模块和注意力机制的交通标志检测算法。该算法在YOLOv5(You Only Look Once version 5)算法的基础上改进,选用感受野模块(Receptive Field Block,RFB)替换原骨干网络中的空间金字塔池化(Spatial Pyramid Pooling,SPP)模块,在特征融合网络中嵌入高效通道注意模块(Efficient Channel Attention Module,ECAM)和卷积块注意模块(Convolutional Block Attention Module,CBAM),选用矩阵非极大值抑制(Matrix Non-Maximum Suppression,Matrix NMS)筛选候选框以提升算法的检测精度和检测速度。实验结果表明,在模型参数量与原网络相比未变化的前提下,该算法的均值平均精度达到了82.31%,与原算法相比提升了8.59%,检测速度达到了51.89 frame·s^(-1),且该算法在各个测试场景中未出现错检漏检现象,证明其泛化能力优于原算法,可以实时检测交通标志。
文摘为提高智能网联(connected and automated,CA)卡车、小车及人工驾驶卡车、小车的混合流道路通行能力,提出基于排强度和渗透率的CA车辆单独编队和合作编队策略.分别设计两种策略下混合流车辆跟驰模式,推导出基于改进Markov模型,涵盖CA车辆渗透率和排强度的车辆状态转移概率;分析两种策略下CA车辆队列分布,建立各策略下的混合流道路容量模型,并通过理论证明和仿真实验予以验证.结果表明,与不编队策略相比,两种策略下道路容量分别提高1.23%~49.62%和1.47%~60.34%,合作编队策略与单独编队策略相比能将道路容量再提高11%;当CA车辆渗透率大于50%和排强度大于0时,编队策略对道路容量的提升效果更显著,容量能提高13.27%~60.34%;单独编队策略下CA小车和CA卡车最大队列规模分别为8辆和6辆,合作编队下CA车辆最大队列规模为8辆.
文摘The modeling of network traffic is important for the design and application of networks, but little is known as to the characteristics of distribution of packets in network traffic. In this letter the distribution of packets in network traffic is explored.
基金The Cultivation Fund of the Key Scientif-ic and Technical Innovation Project,Ministry of Education of China
文摘To measure the length of traffic queue, a vehicle motion model at intersections was built, and based on it the effective traffic queue was def＇med. Color images segmentation and frame differencing technique were used to detect the foreground and the moving vehicles by detecting regions of the images, and then measure the length of effective traffic queue. By processing the image sequence acquired at certain intersection, the results prove that it is able to work out the traffic queue effectively by using the two techniques.
文摘With the rapid development of highway transportation in China,intelligent transportation has become an important part of the traffic structure,and wireless sensor networks are also widely used in intelligent transportation.However,in the wireless traffic sensor network,there is a certain error in the positioning of the anchor blind nodes.In the process of tracking the feedback information,the results of determining the position are very different.Based on the maximum degree of tension,the road traffic wireless Research and analysis of blind node location in sensor networks,and propose solutions and measures to reduce monitoring results.
