文献类型：期刊文献

英文题名：Automatic vehicle speed estimation method for unmanned aerial vehicle images

作者：Long, Hao[1,2]; Chung, Yi-Nung[3]; Li, Jun-De[3]

第一作者：龙浩;Long, Hao

通讯作者：Chung, Yi-Nung

机构：[1] College of Robotics, Department of Electrical Engineering, Beijing Union University, No.97 Beisihuan East Road, Chao Yang District, Beijing, China; [2] National Changhua University of Education, No.2, Shi-Da Road, Changhua, Taiwan; [3] Department of Electrical Engineering, National Changhua University of Education, No. 2, Shi-Da Road, Changhua, Taiwan

第一机构：北京联合大学机器人学院

年份：2018

卷号：9

期号：2

起止页码：442-451

外文期刊名：Journal of Information Hiding and Multimedia Signal Processing

收录：EI(收录号：20181004879181);Scopus(收录号：2-s2.0-85042861680)

基金：Acknowledgment. This work is supported by the Science and Technique Program of Beijing Municipal Education Commission (KM201711417009), the Ministry of Science and Technology under Grant MOST 105-2221-E-018-023, and Applying image processing technology to vehicle seat analysisAY106043.

语种：英文

外文关键词：Aircraft detection - Antennas - Color - Intelligent systems - Road vehicles - Roads and streets - Speed - Unmanned aerial vehicles (UAV)

摘要：This paper presents a solution to solve the detection and estimating the speed problem for road vehicles in images acquired by means of unmanned aerial vehicles (UAVs). UAV photographing which can conveniently take vehicles picture and save time for setting up fixed cameras have become an important part of the intelligent transportation system (ITS). UAV images are characterized by a very high spatial resolution (order of a few centimeters), and consequently by an extremely high level of details which call for appropriate automatic analysis methods. The proposed method starts with the hue, saturation, and value (HSV) color space transformation in order to reduce the influence of light change, and uses the saturation features to remove shadows in the aerial images. Then, it performs a temporal difference process to separate moving objects (vehicles) and backgrounds. The last step of our method is focused on finding out the centroid of vehicles and the moving distance expressed in pixels, and in the end the road lane is used as a scale to estimate the speed of vehicles. The experimental results show that this method performs well in 20 meters to 40 meters in height, and the vehicle speed calculation error is less than 3.47%. ? 2018, Ubiquitous International. All rights reserved.