Nowadays, in the developed countries, the percentage of the elderly is growing. This situation is a consequence of improvements in people's quality life and developments in the medical field. Because of ageing, people have higher probability to be affected by age-related diseases classified in three main groups physical, perceptual and mental. Therefore, the direct consequence is a growing of healthcare system costs and a not negligible financial sustainability issue which the EU will have to face in the next years. One possible solution to tackle this challenge is exploiting the advantages provided by the technology. This paradigm is called Ambient Assisted Living (AAL) and concerns different areas, such as mobility support, health and care, privacy and security, social environment and communication. In this thesis, two different type of sensors will be used to show the potentialities of the technology in the AAL scenario. RGB-Depth cameras and wearable devices will be studied to design affordable solutions. The first one is a fall detection system that uses the distance information between the target and the camera to monitor people inside the covered area. The application will trigger an alarm when recognizes a fall. An alternative implementation of the same solution synchronizes the information provided by a depth camera and a wearable device to classify the activities performed by the user in two groups: Activity Daily Living and fall. In order to assess the fall risk in the elderly, the second proposed application uses the previous sensors configuration to measure kinematic parameters of the body during a specific assessment test called Timed Up and Go. Finally, the third application monitor's the user's movements during an intake activity. Especially, the drinking gesture can be recognized by the system using the depth information to track the hand movements whereas the RGB stream is exploited to classify important objects placed on a table.
Nei paesi sviluppati, la percentuale delle persone anziane è in costante crescita. Questa condizione è dovuta ai risultati raggiunti nel capo medico e nel miglioramento della qualità della vita. Con l'avanzare dell'età, le persone sono più soggette a malattie correlate con l'invecchiamento. Esse sono classificabili in tre gruppi: fisiche, sensoriali e mentali. Come diretta conseguenza dell'aumento della popolazione anziana ci sarà quindi una crescita dei costi nel sistema sanitario, che dovrà essere affrontata dalla UE nei prossimi anni. Una possibile soluzione a questa sfida è l'utilizzo della tecnologia. Questo concetto è chiamato Ambient Assisted living (AAL) e copre diverse aree quali ad esempio il supporto alla mobilità, la cura delle persone, la privacy, la sicurezza e le interazioni sociali. In questa tesi differenti sensori saranno utilizzati per mostrare, attraverso diverse applicazioni, le potenzialità della tecnologia nel contesto dell'AAL. In particolare verranno utilizzate le telecamere RGB-profondità e sensori indossabili. La prima applicazione sfrutta una telecamera di profondità per monitorare la distanza sensore-persona al fine di individuare possibili cadute. Un'implementazione alternativa usa l'informazione di profondità sincronizzata con l'accelerazione fornita da un dispositivo indossabile per classificare le attività realizzate dalla persona in due gruppi: Activity Daily Living e cadute. Al fine di valutare il fattore di rischio caduta negli anziani, la seconda applicazione usa la stessa configurazione descritta in precedenza per misurare i parametri cinematici del corpo durante un test clinico chiamato Timed Up and Go. Infine, la terza applicazione monitora i movimenti della persona durante il pasto per valutare se il soggetto sta seguendo una dieta corretta. L'informazione di profondità viene sfruttata per riconoscere particolari azioni mentre quella RGB per classificare oggetti di interesse come bicchieri o piatti presenti sul tavolo.
Activity monitoring and behaviour analysis using RGB-depth sensors and wearable devices for ambient assisted living applications / Gasparrini, Samuele. - (2016 Feb 26).
Activity monitoring and behaviour analysis using RGB-depth sensors and wearable devices for ambient assisted living applications
Gasparrini, Samuele
2016-02-26
Abstract
Nowadays, in the developed countries, the percentage of the elderly is growing. This situation is a consequence of improvements in people's quality life and developments in the medical field. Because of ageing, people have higher probability to be affected by age-related diseases classified in three main groups physical, perceptual and mental. Therefore, the direct consequence is a growing of healthcare system costs and a not negligible financial sustainability issue which the EU will have to face in the next years. One possible solution to tackle this challenge is exploiting the advantages provided by the technology. This paradigm is called Ambient Assisted Living (AAL) and concerns different areas, such as mobility support, health and care, privacy and security, social environment and communication. In this thesis, two different type of sensors will be used to show the potentialities of the technology in the AAL scenario. RGB-Depth cameras and wearable devices will be studied to design affordable solutions. The first one is a fall detection system that uses the distance information between the target and the camera to monitor people inside the covered area. The application will trigger an alarm when recognizes a fall. An alternative implementation of the same solution synchronizes the information provided by a depth camera and a wearable device to classify the activities performed by the user in two groups: Activity Daily Living and fall. In order to assess the fall risk in the elderly, the second proposed application uses the previous sensors configuration to measure kinematic parameters of the body during a specific assessment test called Timed Up and Go. Finally, the third application monitor's the user's movements during an intake activity. Especially, the drinking gesture can be recognized by the system using the depth information to track the hand movements whereas the RGB stream is exploited to classify important objects placed on a table.File | Dimensione | Formato | |
---|---|---|---|
Tesi_Gasparrini.pdf
accesso aperto
Descrizione: Tesi_Gasparrini.pdf
Tipologia:
Tesi di dottorato
Licenza d'uso:
Creative commons
Dimensione
16.09 MB
Formato
Adobe PDF
|
16.09 MB | Adobe PDF | Visualizza/Apri |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.