A Brief Review on the Validity and Reliability of Microsoft Kinect Sensors for Functional Assessment Applications

Abstract

Kinect sensors are Human Computer Interaction devices oriented to entertainment, but have rapidly spread to several fields such as health care, physical therapy, and training. Their multiple advantages place them at present in a competitive situation compared to traditional solutions. On the other hand, their accuracy and precision for sensitive human applications are still under critical examination. This paper presents a brief literature review on the validity and reliability of the first and the second generation Kinect sensors to get an idea of the feasibility of their propagation as measuring devices in functional assessment applications. Results are difficult to compare because they depend largely on the type of measured elements, the angle of view of the measurement, the distance to the sensor, and even the diversity of human motion features. Nonetheless, they suggest that Kinect sensors are capable of properly identifying posture and motion, but not body or joint rotations, unusual postures, or occlusions.

Keywords

• computer vision
• human computer interaction
• pervasive computing
• reviews
• statistical analysis