Resident identification using kinect depth image data and fuzzy clustering techniques.

Area of Science:

• Gerontology
• Computer Vision
• Biomedical Engineering

Background:

• Passive fall risk assessment is crucial for elderly care in home environments.
• Gait analysis provides valuable insights into mobility and potential fall risks.
• Distinguishing residents from non-residents is essential for accurate data collection in shared living facilities.

Purpose of the Study:

• To present a novel method for identifying older residents using gait information from a single depth camera.
• To enable the collection of resident-exclusive gait data for fall risk analysis.
• To support researchers and health professionals in monitoring elderly individuals' mobility patterns.

Main Methods:

• Continuous depth image acquisition over eight months in a senior housing facility.
• Extraction of shape descriptors (bounding box, image moments) from depth image silhouettes.
• Clustering of extracted features using Possibilistic C Means for resident identification.

Main Results:

• Successful identification of older residents based on extracted gait features.
• Demonstrated capability to filter out data from non-residents.
• Enabled the isolation of gait data specific to the elderly residents.

Conclusions:

• The developed method effectively identifies older residents in home environments using depth camera gait data.
• This technology facilitates passive, continuous monitoring for fall risk assessment.
• Future applications include detecting gait changes indicative of increased fall risk.