Is ORB depending on image resolution?

Question

I'm trying to use OpenCV to detect and extract ORB features from images.

However, the images I'm getting are not normalized (different size, different resolutions, etc...).

I was wondering if I need to normalize my images before extracting ORB features to be able to match them across images?

I know the feature detection is scale invariant, but I'm not sure about what it means for images resolution (for example, 2 images of the same size, with 1 object close, and far in the other should result in a match, even if they have a different scale on the images, but what if the images don't have the same size?).

Should I adapt the patchSize from ORB based on the image size to have (for example if I have an image of 800px and take a patchSize of 20px, should I take a patchSize of 10px for an image of 400px?).

Thank you.

Update: I tested different algorithms (ORB, SURF and SIFT) with high and low resolution images to see how they behave. In this image, objects are the same size, but image resolution is different:

We can see that SIFT is pretty stable, although it has few features. SURF is also pretty stable in terms of keypoints and feature scale. So My guess is that feature matching between a low res and high res images with SIFT and SURF would work, but ORB has much larger feature in low res, so descriptors won't match those in the high res image.

(Same parameters have been used between high and low res feature extraction).

So my guess is that it would be better to SIFT or SURF if we want to do matching between images with different resolutions.

Jourdain

Although ORB is "scale-invariant", it isn't as robust as SURF/SIFT, as mentioned in the ORB research paper. You will get different matches (in number/position/orientation) on images with different resolution so using patch size might work – Teakwood

Yes. I tried to use SIFT/SURF, but unfortunately, they are patented, so OpenCV is throwing me an error, saying I should recompile it with OPENCV_USE_NONFREE but as I installed it with pip, I can't recompile it. – Jourdain

Well, you can get those too actually, see this link – Teakwood

Yes. Unfortunately, that would require me to compile OpenCV, which I was trying to avoid. But I guess I don't really have a choice. – Jourdain

Is there an OpenCV example code for SURF or SIFT? – Perspective

I wonder if you tested the "minimum resolution" for a picture that can give, say at least 20 good matches? – Chiliad

Old thread, but in the example, I see that SIFT behaves the same than the two others. – Editor

Answer 1

According to OpenCV documentation, ORB also use pyramid to produce multiscale-features. Although details are not clear on this page.
If we look at the ORB paper itself, at section 6.1 it is mentioned that images with five different scales are used. But still we are not sure whether you need to compute descriptors on images with different scale manually, or it is already implemented in OpenCV ORB.
Finally, from source code(line 1063 while I write this answer) we see that images with different resolution is computed for keypoint/descriptor extraction. If you track variables you see that there is a scale factor for ORB class which you can access with getScaleFactor method.

In short, ORB tries to perform matching at different scales itself.

Diaconate

I think the scale factor you refer to is the ratio the image is divided by between scale in the pyramid of gaussians. But if I take the same image with 1600x1600px and 800x800px, I won't get the same features, I think. – Jourdain

That is why I said "tries to perform". You will of course get different key points depending on your resolution. Your method of resizing images is also an important factor (linear, bicubic etc). And if you follow the variables in the source code you see that scale mentioned there is used for resizing image, it calls getScale function which depends on scale factor. – Diaconate