how can I use fuzzy c means clustering and membership function instead of threshold?

Lets say I want to detect a moving object in 20 frames, I will perform difference of frames and apply a threshold to the difference and conclude if the difference is greater than the threshold we detect a moving object(foreground) else background. Here I want to replace that threshold with fuzzy membership function by apply fuzzy c-means clustering by creating two clusters and finally applying a membership function.