Event Detection by Eigenvector Decomposition Using Object and Frame Features

We develop an event detection framework that has two significant advantages over past work. First, we introduce an extended set of time-wise and object-wise statistical features including not only the trajectory coordinates but also the histograms and HMM based representations of object's speed, orientation, location, size, and aspect ratio. These features enable detection of events that cannot be detected with the existing trajectory features reported so far. Second, we introduce a spectral clustering algorithm that can automatically estimate the optimal number of clusters. First, we construct feature-wise affinity matrices from the pair-wise similarity scores of objects using the extended set of features. To determine the usual events, we apply eigen-vector decomposition and obtain object clusters. We show that the number of eigenvectors used in the decomposition is proportional to the optimal number of clusters. Unlike the conventional approaches that try to fit predefined models to events, we analyze the conformity of objects using affinity matrices to find the unusual events. We improve the feature selection process by incorporating feature variances. We prove that the clustering stage is not adversely affected by high dimensionality of data space. Our simulations with synthetic and real data reveal that the proposed detection methods accurately detect usual and unusual events.