1
$\begingroup$

Suppose I have a cheap camera that takes 1280x720 photos which are a little bit blurry.

Now suppose I stand somewhere and take 100 photos of the same scene, more or less from the same spot.

Is it possible to combine the data from these images to produce a sharp 1920x1080 image?

(Please note, I'm not asking for a "give me the algorithm" scenario, I mostly want to understand from a theoretical point of view whether or not it is possible to combine poorly sampled data in this way to get higher quality data)

$\endgroup$
0

2 Answers 2

2
$\begingroup$

If the images are somewhat blurry, they all have exactly the same information, combining them doesn’t give you more information.

But if the sensor has fewer pixels than the lens requires (ie it undersamples the image, introducing aliasing), then each image will have slightly different information and can be combined to produce an image with a higher resolution. This is commonly used with IR imagers, which have a low pixel density.

$\endgroup$
0
$\begingroup$

See eg NeRF: https://www.matthewtancik.com/nerf

For combining large number of hand held photos of a static scene relying on known camera movement (IMU). Various derivatives have used this as a starting point for eg super resolution I believe

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.