Title :
Automatic human motion classification from Doppler spectrograms
Author :
Tivive, Fok Ring Chi ; Bouzerdoum, Abdesselam ; Amin, Moeness G.
Author_Institution :
Sch. of Electr., Comput. & Telecommun. Eng., Univ. of Wollongong, Wollongong, NSW, Australia
Abstract :
A technique, recently introduced for visual pattern classification, is successfully applied for classification of human gait based on radar Doppler signatures depicted in the time-frequency domain. It is shown that the proposed classification technique implements steps that, in essence, act on revealing the distinctive Doppler features of the human walking and, as such, allows effective discrimination of various types of human motions characterized by the nature of arm swings. We specifically consider three types of arm motions, namely, free swings, one-arm confined swings, and no-arm swings. The last two arm motions can be indicative of a human carrying objects or a person in stressed situations. The paper explains the different processing stages of motion classification architecture and demonstrates their contributions to the final decision.
Keywords :
Doppler measurement; biomedical ultrasonics; gait analysis; image classification; image motion analysis; medical image processing; Doppler spectrograms; arm motions; automatic human motion classification; free swings; human carrying objects; motion classification architecture; no-arm swings; one-arm confined swings; stressed situations;
Conference_Titel :
Cognitive Information Processing (CIP), 2010 2nd International Workshop on
Conference_Location :
Elba
Print_ISBN :
978-1-4244-6457-9
DOI :
10.1109/CIP.2010.5604253