Microsoft Kinect Sensor Skeleton Tracking Information Essay Example

Kinect is a Microsoft motion sensing device which uses depth map to infer body positions. Using speckle patterns of laser lights from the device’s 3D depth sensors, the depth map is constructed. To analyze the speckle patterns, the Kinect uses a method called structured light which projects light onto a surface and the cameras pick up the deformation of that pattern. Besides the structured light technique, Kinect uses depth from focus and depth from stereo to improve its “vision”. In depth from focus technique, Kinects uses the principle that the blurrier an object is, the further away it is. On the other hand, depth from stereo uses the principle that further objects gets shifted less than closer objects if said objects are observed from another angle. After being able to identify objects’ depth, Kinect uses machine learning and over one hundred thousand depth images of known skeletons to create a randomized decision forest to map depth images to body parts. In the process of learning and creating decision forests, Kinects calculates the following information to understand whether information G is useful:  G\left(\phi\right)=H\left(Q\right)- \sum_{s\in\left\{l,r\right\}}{\frac{\left|Q_s\left(\phi\right)\right|}{\left|Q\right|}H\left(Q_s\left(\phi\right)\right)} (with H is entropy). Using mean shift algorithm, Kinects can compute modes of probability distributions very effectively.