Ask Your Question
0

Image Alignment

asked 2017-02-09 07:46:09 -0500

Alemao gravatar image

I have two aerial photos, a RGB and a InfraRed, this photos are without lens distortion, and are the same size and with the same field of view (99%). I need to make a 4 band image, so i detect some features using ORB, find a homography and use warpperspective to align the photos, but its not good enough, i found some offsets. The photos are big, the size is 11608 x 8708, so when i use SIFT or SURF, my desktop freeze or i have problems with memory. I tried to use FlannMatcher, use BFMatcher, use a mask with photos and used SIFT/SURF, and still not good enough. I know that with this images i can take a perfectly 4 band images because i used another app and it was great, but i need to know how to do this with Opencv. What i can do ?

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted
0

answered 2017-02-10 00:06:52 -0500

Tetragramm gravatar image

IR Visible image registration is an active area of research and there is no best way to do it.

Take a look at current papers and see if anything looks like it will work for you.

edit flag offensive delete link more

Comments

In fact it is a NIR, near-infrared..

Alemao gravatar imageAlemao ( 2017-02-10 04:41:26 -0500 )edit

That's slightly easier, but still not a guarantee. Honestly, multi-scale optical flow might well be better than key points for this.

Tetragramm gravatar imageTetragramm ( 2017-02-10 17:12:22 -0500 )edit

I have done a lot of registration for medical images. I'd like to take a look at the images (scaled down, otherwise too big) to see if I can come up with any idea to help.

Jason - Like Imaging gravatar imageJason - Like Imaging ( 2017-02-12 22:49:02 -0500 )edit

Thanks for the answers!! I will take a look about Multi Scale Optical! If someone wanna see the images,here is a link to download: https://we.tl/k0A1JXdxji

Alemao gravatar imageAlemao ( 2017-02-13 05:59:14 -0500 )edit

Thanks for uploading the images. Here is my procedure: 1. Convert the original images into a kind of feature images (I can send you the processed images). 2. The shift+rotation become more evident in feature images. 3. Use template matching plus certain variation minimization to find the best location.

From my experience with other types of image registration, this procedure would give you pixel-level accuracy.

Jason - Like Imaging gravatar imageJason - Like Imaging ( 2017-02-14 19:51:03 -0500 )edit

How can i convert the images into a "feature image" ? Sorry, i dont know what is a feature image..

Alemao gravatar imageAlemao ( 2017-02-15 06:33:26 -0500 )edit

My email: jshyou@gmail.com. I can send you a pair of feature images to see if they are helpful to you.

Jason - Like Imaging gravatar imageJason - Like Imaging ( 2017-02-15 13:30:54 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2017-02-09 07:46:09 -0500

Seen: 924 times

Last updated: Feb 10 '17