Research On Road Video Information Analysis System Based On Cloud Computing

With the development of China's economy and technology,the traffic volume on the road has increased greatly.The traditional manual methods for monitoring vehicles on the road have been unable to meet the demand.In response to this problem,this paper combines the vehicle identification technology and the cloud computing technology to process the statistical models on the road in real time through streaming video transmission,and publish the results on the webpage to facilitate relevant departments and the public to obtain road information.The experimental results show that the system can effectively process the monitoring video information on the road in real time,and the speed and accuracy of the vehicle identification also meet the design specifications,and the overall operation is good.The main research contents of this paper are as follows:(1)Streaming Media Video Transmission Technology Protocol and Implementation.In order to realize the real-time transmission of road monitoring video to the processing platform,this paper uses the technology of streaming media transmission,realizes the collection of video information at the acquisition end through DirectShow technology,and realizes the encoding of video images through H.264 encoding,and finally,the encoded video information is forwarded by the streaming media server Nginx-rtmp and the streaming media transmission protocol RTMP,the master control platform receives and stores the forwarded information.(2)Vehicle identification algorithm for vehicle images.A new algorithm using multi-scale fusion features is used to classify the vehicle models.Specifically,in the multi-scale case,the Hu moment and LBP features of the image are acquired,and the two features are combined as the fusion features to identify the vehicle.(3)The construction of the cloud computing Hadoop platform.This paper studies the construction of Hadoop platform,constructs the structure design of MapReduce function in cloud computing,combines vehicle identification and cloud computing effectively,improves the processing speed of vehicle image,and satisfies the design index of video image recognition speed.