行人再识别旨在无重叠的视频监控网络中，对于摄像机视场中给定的某个行人，判断其是否出现在其他摄像机视场中的过程，是比行人检测更具挑战性的计算机视觉前沿研究方向。目前大部分研究中行人再识别主要基于封闭式、视角单一建模，通常只能给出确定性的结果，无法满足开放式行人再识别结果的灵活性和多样性，也不能解决行人样本层出不穷和模型更新问题。因此，本课题开展大数据环境下基于深度哈希的开放式行人再识别方法研究。主要研究内容是：视角不变性特征描述子的设计；基于视觉注意机制的行人自动标注算法；大规模视频监控下基于深度哈希的行人检索识别算法研究；大数据环境下的开放式行人再识别模型研究。本课题有望解决行人再识别研究中存在的上述问题，丰富大数据环境下的行人再识别的理论和算法，为计算机视觉的发展提供理论和技术支持。

Person re-identification, identifying the same person’s images in an existing database that come from non-overlapping camera views, is a valuable but challenging task. It is an active and challenging research area in computer vision,including pedestrian detection and localization. The methods of person re-identification often based on closed-end and single visual angle model, result in a fix result. Therefore these method neither can deal with the flexibility and diversity of the result of open-style method, nor can handle the newly arrived samples of pedestrian accompanied with the update of model. Thus, research on open-style method of person re-identification based on deep hashing in big data environment is proposed. The contents of this paper mostly research：(1) Design view-invariant feature descriptor, (2) Pedestrian automatic annotation algorithm based on visual attention mechanism, (3) Pedestrian retrieve and recognition research based on deep hashing algorithm under the large-scale video surveillance. (4) Research on open-style method of person re-identification in big data environment. This project proposal is expected to solve these problems in person re-identification, enrich the theory of person re-identification in big data environment, and promote the development of computer vision.