Python OpenCV template help to implement match filtering methods - expert please

I'm looking for someone who can help to me filter templates and in general improve my code... Currently my application works like below: 1) generate descriptors and key-points for each template, keep in database 2) when starting app, marge descriptors (vstack) and keypoints (extend) by filename (starts from same file-name mean same template) 3) load to memory template images and vstack all descriptors 4) perform knnMatch and simple filter matches 5) if match in index image +1 I need you to help me implement those methods: a) Perform matching in two directions i.e for each point in first image find the best match in second image and vice versa . b) Perform ratio test(ratio test of euclidean distances) between matches to eliminate ambiguous matches c) Perform RANSAC test: it's a model fitting algorithm which finds the best data which fits the model and removes outliers. d) Perform homography: it is an image projection algorithm. The current problem is that sometimes the application returns incorrect results, that is, the returned template is incorrect.