Design And Implementation Of Traffic Video Surveillance Incident System Based On Hadoop

With the rapid development of economy and the speed increase of urbanization construction, the traffic incident have increased year by year because of the traffic vehicle’s increase. The intelligent traffic video surveillance become significant component of urban traffic management.At the same time,the traffic surveillance video data also become the big data. The traditional method to solve the problem of traffic video storage processing and retrieval can not meet the demands of traffic video analysis. How find traffic incident from traffic video and manage them become the urgent problem that need to solve.The paper research to find traffic incident from mass traffic video and manage these data effectively. First, the paper analyze detailedly Hadoop distributed processing model,Hadoop distributed file system,the distributed database Hbase. Secondly,we design the system structure model.According to the feature traffic incident,we accomplish the design and code implementation of every model, including the video upload from client to HDFS,the three algorithm implementation of illegal to drive through red light,drive converse running,illegal parking based on distributed computation, video meta data generation way of onetime video processing and video metadata secondrary analysis.Based on this,the paper make use of HBase to manage meta data and implement the storage and filter of video meta data.Finally, we analyse the experiment result. The result indicates that processing traffic video surveillance based Hadoop is more efficient and faster than traditional processing way. The namagement of video meta data also can let user find it quickly.