With the wide use of photographic surveillance equipment, representation and retrieval of person in video is facing new opportunities and challenges, which is an important problem in the field of intelligent video surveillance. Person representation and retrieval is the key technology of intelligent video surveillance. In this thesis, we study about structured representation of video person and retrieval technology based on Hadoop, which has important practical value.Video person representation and retrieval includes target detection, target tracking, feature extraction, feature matching and retrieval methods. First, we do some analysis about related technology of video retrieval. We propose a person structured representation model for surveillance video and a weighted matching algorithm using person features. We design and implements a distributed video person retrieval method by using Hadoop framework. Experiment results show that the method proposed in this thesis works on a high efficiency, which means it has reference value on solving person retrieval problem of video big data. |