All geek questions in one place

How to use SIFT to compare images

I recently came across a SIFT implementation for C # . I thought it would be great to play with him so that I did.

The implementation generates a set of "percentage points" for any given image. How could I use this information to compare two images?

What I need is the only "meaning of similarity." Can this be formed from two sets of points of interest of two images?

+6
source share
2 answers

You need to run SIFT on both images to get points of interest (let's call them Keypoints) on both images.

After that, you need to find matches between the key points in both images. For this purpose, algorithms are implemented in OpenCV.

The similarity value can be calculated from the number of matches. You might think that if you get more than 4 points, the images will be the same, and you can also calculate the relative rotation between them.

+5
source

You can use the number of matches as an indicator of similarity.

0
source

Source: https://habr.com/ru/post/922212/

More articles:


All Articles