Generate similarity score in percentage from SIFT using opencv

I have been trying to find a way to generate similarity score ( in %) after comparing two images using SIFT in python (2.7.x) opencv (2.4.9). I was only able to find examples that draw lines between matches. How do I proceed with this.

1 Answers

Answers 1

There is an opencv equivalent of vl_ubcmatch function in Matlab.

Here is the excerpt from opencv documentation.

# create BFMatcher object bf = cv2.BFMatcher(cv2.NORM_HAMMING, crossCheck=True)  # Match descriptors. matches = bf.match(des1,des2) 

matches = bf. match (des1, des2) matches the two sets of descriptors and returns a list of DMatch objects. This DMatch object has four attributes: distance, trainIdx, queryIdx, imgIdx. These return values are equivalent of vl_ubcmatch function.

I hope you will find it helpful.

Read More

Proper approach to feature detection with opencv

My goal is to find known logos in static image and videos. I want to achieve that by using feature detection with KAZE or AKAZE and RanSac.

I am aiming for a similar result to: https://www.youtube.com/watch?v=nzrqH...

While experimenting with the detection example from the docs which is great btw, i was facing several issues:

• Object resolution: Differences in size between the known object and the resolution of the scene where the object should be located sometimes breaks the detection algorithm - the object won't be recognized in images with a low resolution although the image quality is still allright for a human eye.
• Color contrast with the background: It seems, that the detection can easily be distracted by different background contrasts (eg: object is logo black on white background, logo in scene is white on black background). How can I make the detection more robust against different luminations and background contrasts?
• Preprocessing: Should there be done any kind of preprocessing of the object / scene? For example enlarge the scene up to a specific size? Is there any guideline how to approach the feature detection in several steps to get the best results?

1 Answers

Answers 1

I think your issue is more complicated than feature-descriptor-matching-homography process. It is more likely oriented to pattern recognition or classification.

You can check this extended paper review of shape matching:

http://www.staff.science.uu.nl/~kreve101/asci/vir2001.pdf

Firstly, the resolution of images is very important, because usually matching operation makes a pixel intensity cross-correlation between your sample image (logo) and your process image, so you will get the best-crosscorrelated area.

In the same way, the background colour intensity is very important because background illumination could affect severally to your final result.

Feature-based methods are widely researched:

http://docs.opencv.org/2.4/modules/features2d/doc/feature_detection_and_description.html

http://docs.opencv.org/2.4/modules/features2d/doc/common_interfaces_of_descriptor_extractors.html

So for example, you can try alternative methods such as:

Hog descritors: Histogram oriented gradients: https://en.wikipedia.org/wiki/Histogram_of_oriented_gradients

Pattern matching or template matching http://docs.opencv.org/2.4/doc/tutorials/imgproc/histograms/template_matching/template_matching.html

I think the lastest (Pattern matching) is the easiest to check your algorithm.

Hope these references helps.

Cheers.

Unai.

Read More

How to use opencv feature matching for detecting copy-move forgery

In my opencv project, I want to detect copy-move forgery in an image. I know how to use the opencv FLANN for feature matching in 2 different image, but I am become so confused on how to use FLANN for detection copy-move forgery in an image.

P.S1: I get the sift keypoints and descriptors of image and stuck in using the feature matching class.

P.S2: the type of feature matching is not important for me.

Thanks in advance.

Update :

These pictures is an example of what I need

And There is a code which matches features of two images and do something like it on two images (not a single one), the code in android native opencv format is like below:

    vector<KeyPoint> keypoints;         Mat descriptors;          // Create a SIFT keypoint detector.         SiftFeatureDetector detector;         detector.detect(image_gray, keypoints);         LOGI("Detected %d Keypoints ...", (int) keypoints.size());          // Compute feature description.         detector.compute(image, keypoints, descriptors);         LOGI("Compute Feature ...");           FlannBasedMatcher matcher;         std::vector< DMatch > matches;         matcher.match( descriptors, descriptors, matches );          double max_dist = 0; double min_dist = 100;          //-- Quick calculation of max and min distances between keypoints           for( int i = 0; i < descriptors.rows; i++ )           { double dist = matches[i].distance;             if( dist < min_dist ) min_dist = dist;             if( dist > max_dist ) max_dist = dist;           }            printf("-- Max dist : %f \n", max_dist );           printf("-- Min dist : %f \n", min_dist );            //-- Draw only "good" matches (i.e. whose distance is less than 2*min_dist,           //-- or a small arbitary value ( 0.02 ) in the event that min_dist is very           //-- small)           //-- PS.- radiusMatch can also be used here.           std::vector< DMatch > good_matches;            for( int i = 0; i < descriptors.rows; i++ )           { if( matches[i].distance <= max(2*min_dist, 0.02) )             { good_matches.push_back( matches[i]); }           }            //-- Draw only "good" matches           Mat img_matches;           drawMatches( image, keypoints, image, keypoints,                        good_matches, img_matches, Scalar::all(-1), Scalar::all(-1),                        vector<char>(), DrawMatchesFlags::NOT_DRAW_SINGLE_POINTS );            //-- Show detected matches //          imshow( "Good Matches", img_matches );           imwrite(imgOutFile, img_matches); 

1 Answers

Answers 1

I don't know if it's a good idea to use keypoints for this problem. I'd rather test template matching (using a sliding window on your image as patch). Compared to keypoints, this method has the disadvantage of being sensible to rotation and scale.

If you want to use keypoints, you can :

• find a set of keypoints (SURF, SIFT, or whatever you want),
• compute the matching score with every other keypoints, with the knnMatch function of the Brute Force Matcher (cv::BFMatcher),

• keep matches between distincts points, i.e. points whose distance is greater than zero (or a threshold).

  int nknn = 10; // max number of matches for each keypoint double minDist = 0.5; // distance threshold  // Match each keypoint with every other keypoints cv::BFMatcher matcher(cv::NORM_L2, false); std::vector< std::vector< cv::DMatch > > matches; matcher.knnMatch(descriptors, descriptors, matches, nknn);  // Compute distance and store distant matches std::vector< cv::DMatch > good_matches; for (int i = 0; i < matches.size(); i++) {     for (int j = 0; j < matches[i].size(); j++)     {         double dist = matches[i][j].distance;         if (dist > minDist)             good_matches.push_back(matches[i][j]);     } }  Mat img_matches; drawMatches(image_gray, keypoints, image_gray, keypoints, good_matches, img_matches); 

Read More