C. Castiello, A.M. Fanelli, M.A. Torsello, T. D'Orazio, and P. Spagnolo (Italy)
Human pose-estimation, image processing, skeletonisation procedure.
In this paper, a pose estimation methodology is described which aims at analysing digital images depicting human figures, in order to properly identifying the body struc ture. The developed methodology does not require the em ployment of any a-priori model, but it is based on a pre processing stage that, besides performing a noise removal procedure, realises a skeletonisation process of the human figure represented into the analysed images. A set of nu merical features is extracted from the obtained skeletons and a geometrical analysis is accomplished to complete the pose estimation process. The methodology has been suc cessfully applied in an experimental session involving both synthetic and real-world images.
Important Links:
Go Back
