This filter uses the values of the same pixel at different points in time to filter out temporal noise in an image. The basis for this is the depth image data. It takes into account any motion in a target scene and helps to reduce motion artifacts.
Use the Strength parameter to define how far back the memory of the temporal filter reaches.
How It Works#
This is a recursive filter. The filter is initialized with the depth data measured for the first frame. This initial depth value is updated with the depth value of the subsequent frame by adding the old value weighted by (k / 255) and the new value weighted by (k - 1 / 255). "k" is the value specified by the
Strength parameter. If motion is detected in the image, i.e., if the depth value for a certain pixel changes drastically to the previous value, the values of the corresponding pixels are reset and reinitialized to avoid motion artifacts.
The following formula illustrates how the filter works:
OUT = (k / 255) * STATE + (1 – k / 255) * IN
OUT = result
IN = new frame
STATE = old frame
k = filter strength (value of the
The value you enter for the
Strength parameter is used as the weight factor k in the formula for calculating the strength of the temporal filter. You can set a value between 50 to 255. The higher the value, the further back the filter reaches.