Holt Double Exponential Filter Algorithm Based on Spatial Limiting

In practice, for the Kinect camera is susceptible to the influence of light, distance and acquisition objects, resulting in the skeletal node data acquired by the camera will inevitably introduce some noise, making the node data deviate from the real value of the problem, this paper proposes a spatial limitation based on holt double-exponential filter algorithm for filtering data anomalies. The method first obtains human 3D skeletal data through Kinect, and then carries out spatial limiting processing for each data, when the skeletal node data exceeds the limitation of the amplitude, take the intersection coordinates of the spatial straight line and the sphere with the above momentary smoothing value as the sphere center, and the amplitude as the radius as the valid value of the data; when the skeletal data is within the limitation of the amplitude, it is believed that the data itself is a valid value, and finally the data after the limitation of amplitude is fed into the holt filter to filter the data. Finally, the amplitude-limited data is fed into the holt filter to smooth the data for prediction. The experimental results show that the algorithm in this paper can effectively remove the influence of data noise and provide a good smoothing prediction process for the data.