Compute similarity measure in feature matching (BFMatcher) in OpenCV-CodePudding

I am comparing images and I have used BFMatcher to perform feature matching

My actual code is:

def get_similarity_from_desc(approach, query_desc, corp_desc):
    if approach == 'sift':
        # BFMatcher with euclidean distance
        bf = cv.BFMatcher()
    else:
        # BFMatcher with hamming distance
        bf = cv.BFMatcher(cv.NORM_HAMMING)
    
    matches = bf.knnMatch(query_desc,corp_desc,k=2)
    # Apply ratio test
    good = []
    for m,n in matches:
        if m.distance < 0.75*n.distance:
            good.append([m])
    similarity = ??
    return similarity

I am wondering if it is possible to compute a similarity measure given the list of good matches good and the descriptors of the two images query_desc and corp_desc

At this moment I have thought:

similarity = len(good) / len(matches)

But I think this is not a correct way of determining similarity between two images

Do you know a better approach for computing this measure?

CodePudding user response：

I have finally done this, which seems to work well:

def get_similarity_from_desc(approach, search_desc, idx_desc):
    if approach == 'sift' or approach == 'orb_sift':
        # BFMatcher with euclidean distance
        bf = cv.BFMatcher()
    else:
        # BFMatcher with hamming distance
        bf = cv.BFMatcher(cv.NORM_HAMMING)
    matches = bf.match(search_desc, idx_desc)
    # Distances between search and index features that match
    distances = [m.distance for m in matches]
    # Distance between search and index images
    distance = sum(distances) / len(distances)
    # If distance == 0 -> similarity = 1
    similarity = 1 / (1   distance)
    return similarity

Source: https://linuxtut.com/en/c9497ffb5240622ede01/