摘要
为解决交通监管部门对于高速公路客货混流问题监管效率低效果差的问题,通过视频图像检测法对高速公路进行研究和应用,构建了基于机器学习和计算机视觉的视频图像检测模式,以提高视频检测的稳定性和准确率,提出了基于尺度不变特征变换(scale invariant feature transformation, SIFT)池化的车辆特征提取模型,摒除传统视频背景建模稳定性和准确率不高的缺陷,获取车辆车型特征数据和分道行驶参数,经过试点样本训练后,实验结果表明:车型识别的准确率高达95%以上,车辆分道检测的准确率达到90%左右。
For the mixed flow of passengers and goods on expressways,the traffic supervision department is of low efficiency and bad effect.A video image detection method was used to study and used for the expressway.In order to improve the accuracy and stability of video detection,a video image detection model based on computer vision and depth learning was built then.The scale invariant feature transformation(SIFT)pool based on the scale invariant feature transform was proposed.The model of vehicle feature extraction was used to overcome the shortcomings of traditional video background modeling with low stability and accuracy,and the feature data and running parameters of vehicles were obtained.After the pilot sample training,the results show that the accuracy of vehicle recognition is as high as 95%,and the accuracy of vehicle lane detection can also reach about 90%.
作者
陈钊正
张善关
杜飞
胡勇
张跃进
CHEN Zhao-zheng;ZHANG Shan-guan;DU Fei;HU Yong;ZHANG Yue-jin(Jiangxi Provincial Communication Investment Group Co.,Ltd.,Nanchang 330036,China;Jiangxi Huitong Technology Development Co.,Ltd.,Nanchang 330036,China;School of Information Engineering,East China Jiaotong University,Nanchang 330013,China;Jiangxi Transportation Science Research Institute Co.,Ltd.,Nanchang 330200,China)
出处
《科学技术与工程》
北大核心
2021年第9期3682-3688,共7页
Science Technology and Engineering
基金
国家自然科学基金(61105015)
江西省交通运输厅项目(2018X0016)。
关键词
客货分道
视频检测
尺度不变特征变换池化
机器学习
bus freight traffic separation
video detection
scale invariant feature transformation pooling
machine learning
