Came across this interesting paper which does feature matching (SIFT-like features) between images under a probabilistic formulation. The methods starts with all matches as inliers and as iterations progress gets rid of matches. About 120 citations as of May 2017.
Original implementation uses the SURF detectors, however often times it causes issues with patent on the SURF detector. You may refer to my git to get around this by working with ORB features. Just clone my repo and should work out of the box. Only dependence is OpenCV.