Hi,
Happy New Year 2024 :-)

Francois LE COAT writes:
> I recently found that you can compute a "temporal disparity" from two
> successive images. With an image sequence, performing the perspective
> registration of the sequence, you can obtain a disparity measure. That
> means depth with only one camera, though disparity is usually measured
> by binocular vision. This appears in the following demonstration...
>
>     <https://www.youtube.com/watch?v=nvfHlw3gHS0>
>
> A drone is flying between a forest of trees in the French Vosges. Thanks
> to the optical-flow measured on successive images, the "temporal
> disparity" reveals the forest of trees... You take a reference image,
> and the optical-flow is obtained on two rectified images, then the
> reference is changed when inter-correlation drops below 60%. You can
> perceive the relief in depth with a unique camera, over time.
>
> Happy New Year 2021.

A WEB page was made to illustrate the "Temporal Disparity" experiment...

When we want to determine the movement, there are two simplifying
hypotheses. Either the camera is fixed, and the observed scene is
moving. Or the camera moves, and the scene is static. In the general
case, the camera and the scene are moving, and it is necessary to
segment static and dynamic elements of what is observed. In this case
the camera is fixed, and it observes the person who is located in front
of the computer and moving. The goal of the experiment is to reconstruct
the visible relief, by "monocular depth"...

<https://hebergement.universite-paris-saclay.fr/lecoat/demoweb/profondeur.html>

That is to say we obtain the depth (inversely proportional to the
disparity) by measuring the optical-flow, with a single camera, tough
this measurement is classically made by binocular vision (for disparity)

Regards,

--
Dr. François LE COAT
CNRS - Paris - France
<https://hebergement.universite-paris-saclay.fr/lecoat>