Reflection symmetry

edit

Hi there,

I am writing a project in C++ using opencv to implement an algorithm from a paper and one of the first step that I have to do is to take an image in input and find global dominant symmetry axis of the image using Reflection/Bilateral symmetry.

Here, in this image there are steps I have to do, to obtain the dominant symmetry axis:

I' ve already compute the first step, so I found SIFT keypoints and descriptor of image:

  SiftFeatureDetector detector(0.05,5.0);
  SiftDescriptorExtractor extractor(3.0);

  std::vector<KeyPoint> keypoints;
  detector.detect(img, keypoints);

  Mat featureImg;
  drawKeypoints(img, keypoints, featureImg, Scalar::all(-1), DrawMatchesFlags::DEFAULT );

  imshow("Keypoints 1", featureImg);

  Mat descriptor;
  extractor.compute(img, keypoints, descriptor);

Now, I have to find (starting from available SIFT points) pairs of locally symmetric points using a threshold of symmetry magnitude but I'm finding a lot of trouble because there are no examples or ideas to start with.

Luckily, for the implementation of the next steps I am following a paper (page 4-5-6):

link text

that describes mathematically the technique to determine the dominant symmetry axis and I decided to write to you to ask if it's the right way or if perhaps there is a simpler solution to my problem.

Thank you for your helps!