ndk076's profile - activity

Thanks you again! I will try it.

Thanks for your answer berak! Could you please tell more details about "derive some heuristic measure, like the average distance beween descriptors, or such". I really need to know how to do it because what I end up to have is an meaningful number to represent distance between two descriptors.

Can we use the Euclidean distance to determine the similarity between two images

• Detect Keypoint image1, image2 using SUFT
• Compute Descriptor image1, image2 using SUFT
• double dif = norm(des1,des2,L2_norm)----> if dif is small -> can we tell that two images similar?

If yes, so what is the threshold to lead to these two images are similar.

And another question is when I compute descriptor for those images. I see that the size of them is not equal. Ex: des1 is 64x96, des2 is 64x85 So the norm function throw an exception.

So is there any way for us to normalize the descriptor to fixed size in order to using Euclidean distance or not?

And I use OpenCV Java API to do these above stuffs currently.

Thanks for your suggestion. I just have found it http://answers.opencv.org/question/25... , is it good to use to determine good homography when we use a way as I describe in my question, and if so, how can we convert to Java because I don't clearly understand some numbers he use in If condition.

Thanks for your suggestion Guanta.
Do you have good example links to solve my current issues if use "bag of words" ?

I'm doing a demo to find all similar images of user's input image in collections of images.

It is similar to this example http://ramsrigoutham.com/2012/11/22/p... but what I want is return list of images that similar to the user's input image (All images are similar that taken in different angle have to be showed in result lists).

So I'm using OpenCV Java to it as following:

• detectKeypoints using FeatureDetector.FAST;
• computeDescriptors using DescriptorExtractor.SURF;
• fetureMatching using DescriptorMatcher.FLANNBASED;
• Then call Calib3d.findHomography(obj, scene, Calib3d.RANSAC, 3) to find Mat object of inline points.

My question here is how can I check if two input images are similar or not because the return result from findHomography(...) is still a Mat object (not something like boolean value TRUE Or FALSE) And my demo require to run in real-time environment. So algorithms I used (FAST, SUFT, FLANNBASED) is correctly to adapt in real-time environment or not ?.

Thanks!

Great!!! Thanks for your support.

Yes, I use SerializationUtils.serialize(data) to convert to byte[]. But I got Mat data type is not compatible again at : m.put(0,0,data ); when I pass float[] to this fucntion:

MatOfKeyPoint mat = new MatOfkeyPoint(); ...... mat.put(0,0,data);

Thank berak for your answer. But when I used it, I got the other issue.

Mat data type is not compatible at: m.get(0,0, bytes);

Hi, I need to convert MatOfKeyPoint to bytes[] and converse in OpenCV Java.

Does anyone have experiment with it before?

I really need it.!!!!

Thanks!