The Design And Implementation Of Massive Video Retrieval System Based On Hadoop Platform

This issue is derived from science and technology benefiting projects named Smart City which is popularized by 3G-WIFI city safety monitoring and control system of Qiqihar. In recent years, video monitoring system has been an important means of Public Security Bureau to ensure public safety and fight crimes. Besides, it is often seen that Public Security Bureau use video monitoring system to fight crimes. As the information storage means of videos is continual and the search and analysis of video is needed later, we must ensure the video data can be stored for a long time. At the same time, the larger the number of camera is, the more the date will be produced. Nowadays there are many different kinds of cameras. Taking a 1/2 inch of CCD camera as an example, there is 341 MB data in an hour and 8GB data in continuous 24 hours. There are 25000 cameras in Qiqihar and the period of the video data storage is about 15 days. With 2930 TB data was produced, in some special situations, the amount of video data could reach PB level. It became a very important and urgent subject of large-scale intelligent video retrieval system to realize efficiently, accurately and quickly search video information.Based on the analysis of the domestic and international video retrieval system and combined with the actual needs of Qiqihar’ Public Security Bureau about the video research, this paper adopt the technology based on HDFS, Map/Reduce and distributed management to achieve the platform of massive video search. In this way,SIFT algorithm will used to calculate the features of video’ keys, remove redundant video data and store the extractive keys in database. At same time, distributed transcoding and massive video search algorithm of LSH will be applied to distributed programming models. To design an structural description of videos android software,We use 3G or WI-FI to realize the collection of onsite videos, succinct label and transmissions to data center server of anomaly diagram. The video retrieval system based on Hadoop not only solves the problem of storage capacity of massive videos, improve the efficiency and accuracy of retrieval but also help monitoring staff to retrieval video information in real time as well as to find required information as soon as possible.