An Adaptive Dictionary Learning Approach for Modeling Dynamical Textures

摘要

Video representation is an important and challenging task in the computervision community. In this paper, we assume that image frames of a moving scenecan be modeled as a Linear Dynamical System. We propose a sparse codingframework, named adaptive video dictionary learning (AVDL), to model a videoadaptively. The developed framework is able to capture the dynamics of a movingscene by exploring both sparse properties and the temporal correlations ofconsecutive video frames. The proposed method is compared with state of the artvideo processing methods on several benchmark data sequences, which exhibitappearance changes and heavy occlusions.