Dynamic textures for human movement recognition

Human motion can be seen as a type of texture pattern. In this paper, we adopt the ideas of spatiotemporal analysis and the use of local features for motion description. movements with dynamic texture features. The proposed method is computationally simple and suitable for various applications such as action and gait recognition. We use Gentle AdaBoost to perform feature selection and build strong models f classifiers. We verify the performance of our methods on the challenging KTH and USF datasets, achieving high accuracy. Categories and Subject Descriptors I.4.7 [Image Processing and Computer Vision Measurement – Feature representation, Texture; Processing and Computer Vision]: Scene Analysis I.5.2 [Pattern Recognition]: Design Methodology evaluation and selection General Terms Design, Experimentation Keywords Dynamic texture, LBP-TOP, Gentle AdaBoost, K

• Type: Journal
• Year: 2010
• Where: CIVR
• Authors: Vili Kellokumpu, Guoying Zhao, Matti Pietikäinen
• Keywords: CIVR 2010, Computer Vision, Dynamic Texture, Gentle Adaboost, Image Analysis