From: Multiple Kinect based system to monitor and analyze key performance indicators of physical training
References | Number of Kinect sensors used | Data pre-processing (filtering) methods used | Data (skeleton) fusion methods used |
---|---|---|---|
[3] | 3 | Mamdani Fuzzy Inference Scheme (FIS) | Maximization of energy sum function over candidate positions |
[8] | 2 | Not used | Coordinate transform with joint state inference |
[12] | 4 | Not used | Point cloud matching |
[16] | 2 | Not used | Point cloud registration in a maximum a posteriori framework |
[20] | 4 | Information weighted-consensus filter (IWCF) | Interactive multiple model (IMM) |
[31] | 3 | Not used | Iterative Closest Point (ICP) algorithm & Rigid Transformation (RT) to reference frame |
[39] | 5 | Kalman filter | Weighted Measurement Fusion |
[56] | 3 | Kalman filter | Trajectory matching |
[57] | 4 | Median filter | Iterative Closest Point (ICP) algorithm |
[65] | 3 | Double exponential smoothing | Weighted position averaging |