K. Spatial segmentation of temporal texture using mixture linear models.pdf

Spatial Segmentation of Temporal Texture Using Mixture Linear Models Lee Cooper Department of Electrical Engineering Ohio State University Columbus, OH 43210 cooperl@ece.osu.edu Jun Liu Biomedical Engineering Center Ohio State University Columbus, OH 43210, USA liu.314@osu.edu Kun Huang ? Department of Biomedical Informatics Ohio State University Medical Center Columbus, OH 43210, USA khuang@bmi.osu.edu Abstract In this paper we propose a novel approach for the spatial segmentation of video sequences containing mul- tiple temporal textures. This work is based on the notion that a single temporal texture can be represented by a low- dimensional linear model. For scenes containing multiple temporal textures, e.g. trees swaying adjacent a flowing river, we extend the single linear model to a mixture of linear models and segment the scene by identifying sub- spaces within the data using robust generalized principal component analysis (GPCA). Computation is reduced to minutes in Matlab by first identifying models from a sam- pling of the sequence and using the derived models to seg- ment the remaining data. The effectiveness of our method has been demonstrated in several examples including an application in biomedical image analysis. 1 Introduction Modeling motion is a fundamental issue in video analysis and is critical in video representa- tion/compression and motion segmentation problems. In this paper we address a special class of scenes, those that contain multiple instances of so-called temporal texture, described in [11] as texture with motion. Previous works on temporal texture usually focused on synthesis with the aim of generating an artificial video ? This work is partially supported by the startup funding from the Department of Biomedical Informatics, Ohio State University Medical Center. sequence of arbitrary length with perceptual likeness to the original. Prior schemes usually model the temporal texture using either a single stochastic process or dynami- ca