I have to use control points to improve finding the same part of an image in another one.
by using this code, the control points coordinate determined but I want to display matched area in Image2.
Indead I want to crop a part of Image1, then finding that in Image2 by control points.
how can I find coordinate to crop this founded area?
My code:
- scale = 1;
- J = imresize(Image1, scale); % Try varying the scale factor.
- theta = 90;
- distorted = imrotate(J,theta); % Try varying the angle, theta.
- ptsOriginal = detectSURFFeatures(Image2);
- ptsDistorted = detectSURFFeatures(distorted);
- [featuresOriginal, validPtsOriginal] = extractFeatures(Image2, ptsOriginal);
- [featuresDistorted, validPtsDistorted] = extractFeatures(distorted, ptsDistorted);
- indexPairs = matchFeatures(featuresOriginal, featuresDistorted);
- matchedOriginal = validPtsOriginal(indexPairs(:,1));
- matchedDistorted = validPtsDistorted(indexPairs(:,2));
- x1 = matchedDistorted.Location(1,1)
- y1 = matchedDistorted.Location(1,2)
- DisplayMatchArea=imcrop(B,[x1 y1 size(distorted,2)-1 size(distorted,1)-1]);
This is a good question. Actually, it looks like there are two main questions: (1) question about the use of control points in image processing and (2) how to use control points to find a pair of matching image regions.
For the first question, a good place to start is
https://www.researchgate.net/publication/258713398_Automatic_control-point_selection_for_image_registration_using_disparity_fitting
and
https://www.researchgate.net/publication/259569207_Development_and_Status_of_Image_Matching_in_Photogrammetry
More to the point in terms of question (2), see
https://www.researchgate.net/publication/220939509_Integrating_Range_and_Texture_Information_for_3D_Face_Recognition/figures?lo=1
Article Automatic control-point selection for image registration usi...
Article Development and Status of Image Matching in Photogrammetry
Conference Paper Integrating Range and Texture Information for 3D Face Recognition
Dear Fahime
It seems a very nice question. You should go through the resource provided by Professor James. I am adding some resources that may help you regarding this.
Regards
Ehtesham
http://in.mathworks.com/help/images/start-the-control-point-selection-tool.html
http://in.mathworks.com/help/images/select-matching-control-point-pairs.html
http://in.mathworks.com/help/images/control-point-selection-procedure.html
http://in.mathworks.com/help/images/using-correlation-to-improve-control-points.html
https://in.mathworks.com/help/images/ref/cpselect.html
http://www-rohan.sdsu.edu/doc/matlab/toolbox/images/cpcorr.html
Article Moment-based approach to registration of images with affine ...
Article A feature-based image registration algorithm using the multi...
Hey, simply put, you matched points using SURF, those matches contains a number of false match , you need to eliminate those by applying Ransac and a model relating the two scenes. in your case the model is Homography (applied for pure rotation or when the scene is planar).
so apply a Ransac Homography to your matched pairs (OpenCV got it) and you'll have the inliers and the transformation from Image one to Image2
- Badges
- Science topic
- Similar topics
- Philosophy
- Moral Philosophy
- Free Will