I have an image and i have marked the object in the image by a Rect(92,65,105,296). I then calculated the gaussian mixture models for both the region inside the rect and the region outside the rect using cv::EM. Now i want to calculate the following probabilities, i can't seem to understand how they could be calculated from the cv::EM results.
For each pixel x of the image, find the probability:
P(x|Ggd) / (P(x|Bgd) + P(x|Fgd))
I would appreciate any help on how these could be calculated.