Find transform between two cameras

asked 2017-03-21 10:19:21 -0500

giralto gravatar image

updated 2017-03-21 10:53:41 -0500

I have two camera of known intrinsic and extrinsic parameters. The parameters are in the same coordinate space and of same units.

Using Python 3, I would like to transform an image taken with one camera to the other camera's viewpoint so that I can overlay them. An example of what I'd like to achieve is: http://www.buildingconservation.com/articles/thermal-imaging/pic_1.jpg (www.buildingconservation.com/articles...) (Not enough "karma" to include the image/link.)

The cameras are close to each other and pointing at the same direction, but I can't make the assumption that they are aligned on the x or y axis.

So far I have read a lot of documentation about stereo imaging. But as far as I understand, my problem is different.

Am I dealing with an image rectification problem?

Is there a solution that uses the intrinsic and extrinsic matrices?

My setup has fixed, known parameters. Yet, because of other constraints I can't use the method of matching key points from the two images to compute the homography.

edit retag flag offensive close merge delete

Comments

I don't think you are dealing with a rectifcation problem. You are dealing wit a problem of matching and homography and may be resampling if visible and infrared resolution sensor are differents. In your example I think you will need data fusion too.

Are extrinsic parameters in same coordinates space?

LBerger gravatar imageLBerger ( 2017-03-21 10:45:18 -0500 )edit

Yes, the parameters are in the same coordinate space. (I will edit the question to add this information.)

giralto gravatar imagegiralto ( 2017-03-21 10:53:07 -0500 )edit

Finally try with stereorectify.If your object is far (relative to base) disparity will be 0 you can simply add images...If it is not far you have match images

LBerger gravatar imageLBerger ( 2017-03-21 13:02:10 -0500 )edit