Video Anomaly Detection in Spatiotemporal Context (2010)
We propose a context-aware method to detect anomalous video event from surveillance video. By tracking all moving objects in the video, three different levels of spatiotemporal contexts are considered, i.e., point anomaly of a video object, sequential anomaly of an object trajectory, and co-occurrence anomaly of multiple video objects.
Abnormal Event Detection Based on Trajectory Clustering (2009)
The proposed abnormal video event detection method is based on unsupervised clustering of object trajectories. The normal trajectory clusters are identified from clustering results and are modeled by hidden Markov models (HMM). Any trajectory that cannot be explained by those normal models is declared to be abnormal.
Image/Video Retargetting by Anchor Point Sampling and Mapping (2008)
We proposes an efficient approach for image/video resizing while effectively preserving the important contents in the image/video. The combination of sampling and mapping approaches greatly reduces the computational cost yet leads to the globally sound solution.
Detecting Contextual Anomalies of Crowd Motion in Surveillance Video (2008)
We introduce a new concept of contextual anomaly into the field of crowd analysis, i.e., the behaviors themselves are normal but they are anomalous in a specific context. Our system follows an unsupervised approach. It automatically discovers important contextual information from the crowd video and detects the blobs corresponding to contextually anomalous behaviors.
Home Video Structuring with Shot Clustering (2005)
Content-based video structuring above shot level faces technical challenges in semantic feature extraction and flexible shot cluster organization. Aiming at solving these problems, a two-layer shot clustering approach for home video structuring, which operates directly in MPEG domain, is proposed.
News Video Indexing and Abstraction by Specific Visual Cues (2004)
In this project we addresses the tasks of providing the semantic structure and generating the abstraction of content in broadcast news, based on extraction of two specific visual cues, Main Speaker Close-Up (MSC) and news caption.
