Indoor occupant behaviour monitoring with the use of a depth registration camera. (15th January 2019)

• Record Type: Journal Article
• Authors: Dziedzic, Jakub Wladyslaw; Da, Yan; Novakovic, Vojislav
• Is Part Of: Building and environment. Volume 148(2019)
• Page Start: 44
• Page End: 54
• Publication Date: 2019-01-15
• Subjects: Occupant behaviour -- Building energy performance -- Depth registration -- Pattern recognition -- Big data; Buildings -- Environmental engineering -- Periodicals; Building -- Research -- Periodicals; Constructions -- Technique de l'environnement -- Périodiques; Electronic journals; 696
• DOI: 10.1016/j.buildenv.2018.10.032
• Languages: English
• ISSNs: 0360-1323

Abstract

Abstract: Registration, identification, and re-creation of an indoor occupant's actions are challenges in the field of building energy performance. Commonly used measurement technologies are capable of capturing partial information regarding the occupants' activity. However, the combination of all existing inputs cannot grant access to a satisfying description of occupant behaviour that allows capturing profiles of occupants' intentions and habits. It seems that there is a missing type of data that could be used as a connection platform for already existing inputs. To connect existing data sets, there is a need to deploy a monitoring method that can identify particular individuals; however, it must do so while still providing a certain level of privacy among the monitored occupants. Fulfilment of these standards can be achieved through the use of the depth registration technique. The entertainment industry popularized this registration technique, but this registration method has many other applications in the fields of medicine and computer vision. The most commonly used device (Microsoft Kinect) delivers high-frequency sampling (up to 30 Hz) and a moderate measurement range (up to 5 m), which allows its usage in the monitoring of medium-sized indoor spaces. The delivered input data do not allow for the direct identification of the monitored person, and it does not require any interaction from the occupants to initialise the monitoring procedure. Due to these reasons, the potential of this measurement method was explored in terms of becoming an in situ indoor occupant behaviour monitoring technique. Highlights: The high-frequency monitoring of occupant action can grant access to the understatement of the reasons their actions. Depth registration allows to identify occupants indirectly. New monitoring technique enables dynamic measurement of the spatial positioning, estimation of clothing coverage and activity level.