Tips for subtracting background in the face of noise

Question

Subtraction of the background is an important primitive in computer vision. I study the various methods that have been developed, and I began to think about how to perform background subtraction in the face of random, salt and pepper noises.

In a system like Microsoft Kinect, an infrared camera will produce random noise quite consistently. If you try to cross out the background from the depth view, how can you avoid the problem of this random noise while reliably subtracting the background?

Answer 1

as you said, noise and other non-stationary parts of your background can cause segmentation problems, I mean changes in lighting or other moving things in the background.

- , , .

( ) . , , , . , .

, , , , . - Median Filter Morpholocigal Operators () . .

, ...

Answer 2

(cc) , cc, . (, , ) ([e-mail]).

Answer 3

, - . , .

However, I disagree with his suggestion that this happens on a segmented binary image. Median filtering is very low level and should be applied to raw data before any further processing.