Volume-based Video analysis using 3D Segmentation Techniques

Video content understanding for surveillance and security applications such as smart CCTV cameras has become a research hotspot in the last decade. This paper presents applications of volume construction and volume-based cluster segmentation approaches to video event detection. It starts with a description of the translation between original video frames and 3D volume structures denoted by spatial and temporal features. It then highlights the volume array structure, a so called “pre-suspicion” mechanism. The focus of the work is on devising an effective and efficient voxel-based segmentation technique suitable to the volumetric nature of video events through deploying innovative 3D clustering methods. It is supported by the design and experiment on the 3D data compression techniques for accelerating the pre-processing of the original video data. An evaluation on the performance of the developed methods is presented at the end.