Denoising of Joint Tracking Data by Kinect Sensors Using Clustered Gaussian Process Regression

Abstract

Using Kinect sensors to monitor and provide feedback to patients performing intervention or rehabilitation exercises is an upcoming trend in healthcare. However, the joint positions measured by the Kinect sensor are often unreliable, especially for joints that are occluded by other parts of the body. Motion capture (MOCAP) systems using multiple cameras from different view angles can accurately track marker positions on the patient. But such systems are costly and inconvenient to patients. In this work, we simultaneously capture the joint positions using both a Kinect sensor and a MOCAP system during a training stage and train a Gaussian Process regression model to map the noisy Kinect measurements to the more accurate MOCAP measurements. To deal with the inherent variations in limb lengths and body postures among different people, we further propose a joint standardization method, which translates the raw joint positions of different people into a standard coordinate, where the distance between each pair of adjacent joints is kept at a reference distance. Our experiments show that the denoised Kinect measurements by the proposed method are more accurate than several benchmark methods.

Keywords: Gaussian process regression; denoising of Kinect measurements.

Figures

Figure 1

Data capture system set up

Figure 2

Motion capture system marker positions: (a) front upper body, (b) back upper body [5].

Figure 3

Kinect V2 Joint position [7].

Figure 4

Sample upper body skeletons. (a) raw Kinect data and raw MOCAP data. (b) standardized Kinect data and MOCAP data.

Figure 5

(a) Average computation time per time sample with different cluster number, (b) error per joint with different cluster number.

Figure 6

The raw captured skeleton by Kinect (green) and the denoised skeleton by the proposed method (red) overlaid on the RGB image. (a) push-down-pumping exercise; (b) over-the-head-pumping exercise.

Figure 7

Trajectory of the joint during the exercise. (a) Trajectory of left wrist when doing the push-down-pumping exercise. (The red region is when the wrist close to the chest) (b) Trajectory of left elbow when doing the over-head-pumping exercise.

Figure 7

Trajectory of the joint during the exercise. (a) Trajectory of left wrist when doing the push-down-pumping exercise. (The red region is when the wrist close to the chest) (b) Trajectory of left elbow when doing the over-head-pumping exercise.