Questo sito utilizza cookie di terze parti per inviarti pubblicità in linea con le tue preferenze. Se vuoi saperne di più clicca QUI 
Chiudendo questo banner, scorrendo questa pagina, cliccando su un link o proseguendo la navigazione in altra maniera, acconsenti all'uso dei cookie. OK

Un modello e un sistema per l'analisi e la segmentazione del movimento umano. Uno studio sulla danza

L'anteprima di questa tesi è scaricabile in PDF gratuitamente.
Per scaricare il file PDF è necessario essere iscritto a Tesionline.
L'iscrizione non comporta alcun costo. Mostra/Nascondi contenuto.

Introduction The figure depicted in the previous page shows the three areas on which human motion analysis mainly concentrates, according to Aggarwal and Cai’s point of view (1997): body structure analysis (developed mainly by biomechanics), tracking (for example for video conferencing scenarios, but also addressed by multimedia content analysis) and recognition (useful, for example, in the video-surveillance field and also in case of segmentation based on recognition of single gestures). In (Aggarwal and Cai, 1997) the two authors also point out that, talking about motion analysis, there is always a trade-off between feature complexity and tracking efficiency: lower level features, such as points, are easier to extract, but relatively more difficult to track than higher-level features such as blobs and 3D volumes. This has been confirmed by our work: we have concentrated on tracking of points corresponding to joints or to the Centre of Mass of the body and the algorithm employed for motion segmentation needs precise tracking of the chosen points. This is the reason why an accurate manual tracking has sometimes been necessary: to obtain reliable values of position. We can certainly affirm that getting consistent values has been one of the bottlenecks of our approach to segmentation. Another possible way to approach the issue of motion segmentation might be through considering motion recognition: recognizing a certain movement allows its distinction among others in a flow of different actions: therefore we may have segmentation based on recognition of certain moves even if they are meshed with other unrecognisable movements. Two typical approaches to motion recognition are addressed in the publication by Aggarwal and Cai (1997): that based on template matching some given images to pre-stored patterns (the preferred approach by Aggarwal, who used it in (Aggarwal and Ali, 2001)) and that based on a state-space models. To say the truth, we have not used the first approach neither the second, since ours is based on motion segmentation conceived as a step before or even disconnected with motion recognition. We have tried to make motion segmentation without any recognition of the movements. Our approach is rather an attempt to divide streams of dance movements in phases according to some kinematical features, with a particular focus on how observers would execute such a task. In fact, the basic units of movement we can detect (e.g., in a dance) and cluster together can be considered and analysed under three perspectives, sometimes connected one to another: that of the performer (it means that attention is focused on the physical execution of the movement and expressive gestures 4 represent items of acted moves), that of the observer (it is based on the perception of movements) and finally that of the choreographer. This topic is detailed in the paragraph “The dance field”. 4 The concept of expressive gesture is deepened in the paragraph “Experimental psychology”.

Anteprima della Tesi di Elisa Rocca

Anteprima della tesi: Un modello e un sistema per l'analisi e la segmentazione del movimento umano. Uno studio sulla danza, Pagina 6

Tesi di Laurea

Facoltà: Ingegneria

Autore: Elisa Rocca Contatta »

Composta da 138 pagine.


Questa tesi ha raggiunto 3979 click dal 19/05/2004.

Disponibile in PDF, la consultazione è esclusivamente in formato digitale.