Human motion analysis: methodologies and applications

• Type: Article in International Conference Proceedings Book
• Date: 2008

Authors

Maria João M. Vasconcelos (Author) Other; João Manuel R. S. Tavares (Author) FEUP

Conference proceedings International

Title: CMBBE 2008

Initial page: 1pag.

8th International Symposium on Computer Methods in Biomechanics and Biomedical Engineering

Sheraton Hotel, Porto, Portugal, February 27th - March 1st 2008

Scientific classification

CORDIS: Technological sciences > Engineering

Other information

Abstract (EN): The study of motion is one of the most interesting areas in Computational Vision, particularly the human motion research. During these last decades several works were presented regarding this subject. Human motion analysis is complex, non-linear and time variant and its tracking can be done using Computational Vision through, for example, human modeling. In order to make human motion analysis more computational tractable, some assumptions are often made. For instance: regarding to movements involved, if the images have one or more persons in the workspace at the same time, if the cameras are static or not or the subject remains inside the workspace being acquired; regarding the environment, if light conditions are constant, the background is static and uniform or not; and regarding the subject in analysis, if its shape and motion are known, if has special markers or clothes. In this work, we intend to present a review about the computational methodologies used in human motion, their advantages and disadvantages and some of their main applications. The study of human motion in image sequences usually follows a general framework: feature extraction, feature correspondence and high-level processing. Feature extraction is related to human motion modeling; these models can be built using stick figures, 2D contours or volumetric models. In feature correspondence the problem of matching two features between two consecutive image frames is approach; one of the difficulties here is to treat occlusion problems that may occur. After the features are extracted and correctly matched over an image sequence, high-level processing can be used, for instance, in the recognition of human movements or activities. As examples of human motion applications we can refer: human recognition in surveillance systems; motion analysis in clinical studies, for example, on orthopedic patients; biomechanical study of athletes, in order to improve their performances; gait classification.

Language: English

Type (Professor's evaluation): Scientific