Implementation of Linear Kalman Filter on Kinect V2 skeleton tracking data using MATLAB.
Smooth out the noise in Kinect's skeleton tracking algorithm using simple linear Kalman Filtering. Tested on MATLAB R2018a and Microsoft Kinect V2 for Windows.
Simply run the color_kalman_kinectv2.m or the depth_kalman_kinectv2.m file to for the kalman filtering of Kinect V2. The pointcloud.m is simply a way to compute a pointcloud in MATLAB using the Kinect V2 Sensor. Keep in mind that the Kinect V2 is pretty slow in MATLAB when the skeleton tracking is enabled.
Full list of all requirements
Example of Kalman and Kinect Joint Tracking, where the red circle represents the kalman Joint Tracking and the green circle the kinect's joint value:
Same as the Color Example
Plots comparing the kinect's values to the kalman estimation.