In multi-camera video surveillance, it is challenging to represent videosfrom different cameras properly and fuse them efficiently for specificapplications such as human activity recognition and clustering. In this paper,a novel representation for multi-camera video data, namely the ProductGrassmann Manifold (PGM), is proposed to model video sequences as points on theGrassmann manifold and integrate them as a whole in the product manifold form.Additionally, with a new geometry metric on the product manifold, theconventional Low Rank Representation (LRR) model is extended onto PGM and thenew LRR model can be used for clustering non-linear data, such as multi-cameravideo data. To evaluate the proposed method, a number of clustering experimentsare conducted on several multi-camera video datasets of human activity,including Dongzhimen Transport Hub Crowd action dataset, ACT 42 Human actiondataset and SKIG action dataset. The experiment results show that the proposedmethod outperforms many state-of-the-art clustering methods.
展开▼
