Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > sci.image.processing > #4919 > unrolled thread

Optical Inertia

Back to article view | Back to sci.image.processing

Contents

  Optical Inertia Francois LE COAT <lecoat@atari.org> - 2024-02-28 14:45 +0100
    Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2024-03-28 16:15 +0100
      Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2024-04-03 18:45 +0200
        Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2024-10-17 15:15 +0200
          Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2024-10-29 19:00 +0100
            Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2024-11-21 16:35 +0100
              Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2025-02-24 16:30 +0100
                Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2025-04-01 19:15 +0200
                  Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2025-05-15 15:00 +0200
                    Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2025-05-30 15:51 +0200
                      Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2025-06-18 15:15 +0200
                        Re: Optical Inertia Francois LE COAT <lecoat@atari.org> - 2025-09-16 15:30 +0200
                          Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2025-10-21 16:35 +0200
                            Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2025-11-04 16:15 +0100
                              Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2025-11-18 15:10 +0100
                                Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2025-12-09 15:30 +0100
                                  Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2026-01-14 15:15 +0100
                                    Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2026-04-28 15:45 +0200
                                      Re: Optical Inertia Francois LE COAT <lecoat@archimedium.fr> - 2026-07-02 17:00 +0200

#4919 — Optical Inertia

Hi,

The experiment from Hernan Badino was redone. You can see it there...

     <https://www.youtube.com/watch?v=fqWdSfN9FiA> Source

The main interest is that video is looping, and the result is almost:

     <https://www.youtube.com/watch?v=0ZPJmnBh03M> Reworked

Well, Hernan Badino is moving his head when he is walking, so the
reconstructed trajectory is not perfectly looping at the end. But
we can reconstruct the movement almost perfectly. We use OpenCV
for image processing, and POV-Ray for 3D representation. We have
to determine projective dominant motion in the video with a
reference image, and change it when correlation drops below 80%.

We have a 3D inertial model of motion, that's why POV-Ray helps =)

Best regards,

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

[toc] | [next] | [standalone]

#4920

Hi,

The principle of dominant 2D motion appeared at INRIA, it is here:

<https://www.irisa.fr/vista/Themes/Logiciel/Motion-2D/Motion-2D.html>

In our case, the dominant motion estimated from the approximation of
optical-flow (DIS - Dense Inverse Search OpenCV) is 3D and projective.

Francois LE COAT writes:
> The experiment from Hernan Badino was redone. You can see it there...
> 
>      <https://www.youtube.com/watch?v=fqWdSfN9FiA> Source
> 
> The main interest is that video is looping, and the result is almost:
> 
>      <https://www.youtube.com/watch?v=0ZPJmnBh03M> Reworked
> 
> Well, Hernan Badino is moving his head when he is walking, so the
> reconstructed trajectory is not perfectly looping at the end. But
> we can reconstruct the movement almost perfectly. We use OpenCV
> for image processing, and POV-Ray for 3D representation. We have
> to determine projective dominant motion in the video with a
> reference image, and change it when correlation drops below 80%.
> 
> We have a 3D inertial model of motion, that's why POV-Ray helps =)

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4921

Hi,

Francois LE COAT writes:
> The principle of dominant 2D motion appeared at INRIA, it is here:
> 
> <https://www.irisa.fr/vista/Themes/Logiciel/Motion-2D/Motion-2D.html>
> 
> In our case, the dominant motion estimated from the approximation of
> optical-flow (DIS - Dense Inverse Search OpenCV) is 3D and projective.
> 
> Francois LE COAT writes:
>> The experiment from Hernan Badino was redone. You can see it there...
>>
>>      <https://www.youtube.com/watch?v=fqWdSfN9FiA> Source
>>
>> The main interest is that video is looping, and the result is almost:
>>
>>      <https://www.youtube.com/watch?v=0ZPJmnBh03M> Reworked
>>
>> Well, Hernan Badino is moving his head when he is walking, so the
>> reconstructed trajectory is not perfectly looping at the end. But
>> we can reconstruct the movement almost perfectly. We use OpenCV
>> for image processing, and POV-Ray for 3D representation. We have
>> to determine projective dominant motion in the video with a
>> reference image, and change it when correlation drops below 80%.
>>
>> We have a 3D inertial model of motion, that's why POV-Ray helps =)

Three drones are flying between forests of trees. Thanks to the
optical-flow (DIS OpenCV) measured on successive images, the
"temporal disparity" reveals the forest of trees (3rd dimension)...

<https://www.youtube.com/watch?v=QP75EeFVyOI> 1st drone
<https://www.youtube.com/watch?v=fp5Z1Nu4Hko> 2nd drone
<https://www.youtube.com/watch?v=fLxE8iS7fPI> 3rd drone

The interest with the forest is that trajectories are curved, in
order to avoid obstacles. It is measured thanks to a projective
transform, and represented with <Ry,Rz,Tx,Tz> thanks to POV-Ray.
The evolution of the drone is shown in front-view with its camera.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4922

Hi,

Francois LE COAT writes:
>> The principle of dominant 2D motion appeared at INRIA, it is here:
>>
>> <https://www.irisa.fr/vista/Themes/Logiciel/Motion-2D/Motion-2D.html>
>>
>> In our case, the dominant motion estimated from the approximation of
>> optical-flow (DIS - Dense Inverse Search OpenCV) is 3D and projective.
>>
>> Francois LE COAT writes:
>>> The experiment from Hernan Badino was redone. You can see it there...
>>>
>>>      <https://www.youtube.com/watch?v=fqWdSfN9FiA> Source
>>>
>>> The main interest is that video is looping, and the result is almost:
>>>
>>>      <https://www.youtube.com/watch?v=0ZPJmnBh03M> Reworked
>>>
>>> Well, Hernan Badino is moving his head when he is walking, so the
>>> reconstructed trajectory is not perfectly looping at the end. But
>>> we can reconstruct the movement almost perfectly. We use OpenCV
>>> for image processing, and POV-Ray for 3D representation. We have
>>> to determine projective dominant motion in the video with a
>>> reference image, and change it when correlation drops below 80%.
>>>
>>> We have a 3D inertial model of motion, that's why POV-Ray helps =)
> 
> Three drones are flying between forests of trees. Thanks to the
> optical-flow (DIS OpenCV) measured on successive images, the
> "temporal disparity" reveals the forest of trees (3rd dimension)...
> 
> <https://www.youtube.com/watch?v=QP75EeFVyOI> 1st drone
> <https://www.youtube.com/watch?v=fp5Z1Nu4Hko> 2nd drone
> <https://www.youtube.com/watch?v=fLxE8iS7fPI> 3rd drone
> 
> The interest with the forest is that trajectories are curved, in
> order to avoid obstacles. It is measured thanks to a projective
> transform, and represented with <Ry,Rz,Tx,Tz> thanks to POV-Ray.
> The evolution of the drone is shown in front-view with its camera.

It is possible to perceive the relief (in depth) of a scene, when we
have at least two different viewpoints of it. Here is a new example with
a drone flying in the middle of a forest of trees, and from which we
process the video stream from the embedded camera...

<https://www.youtube.com/watch?v=WJ20EBM3PTc>

When the two views of the same scene are distant in space, we speak
of "spatial disparity". In the present case, the two viewpoints are
distant in time, and we then speak of "temporal disparity". This
involves knowing whether the two images of the same scene are acquired
simultaneously, or delayed in time. We can perceive the relief in depth
in this case, with a single camera and its continuous video stream.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4923

Hi,

Francois LE COAT writes:
>>> The principle of dominant 2D motion appeared at INRIA, it is here:
>>>
>>> <https://www.irisa.fr/vista/Themes/Logiciel/Motion-2D/Motion-2D.html>
>>>
>>> In our case, the dominant motion estimated from the approximation of
>>> optical-flow (DIS - Dense Inverse Search OpenCV) is 3D and projective.
>>>
>>>> The experiment from Hernan Badino was redone. You can see it there...
>>>>
>>>>      <https://www.youtube.com/watch?v=fqWdSfN9FiA> Source
>>>>
>>>> The main interest is that video is looping, and the result is almost:
>>>>
>>>>      <https://www.youtube.com/watch?v=0ZPJmnBh03M> Reworked
>>>>
>>>> Well, Hernan Badino is moving his head when he is walking, so the
>>>> reconstructed trajectory is not perfectly looping at the end. But
>>>> we can reconstruct the movement almost perfectly. We use OpenCV
>>>> for image processing, and POV-Ray for 3D representation. We have
>>>> to determine projective dominant motion in the video with a
>>>> reference image, and change it when correlation drops below 80%.
>>>>
>>>> We have a 3D inertial model of motion, that's why POV-Ray helps =)
>>
>> Three drones are flying between forests of trees. Thanks to the
>> optical-flow (DIS OpenCV) measured on successive images, the
>> "temporal disparity" reveals the forest of trees (3rd dimension)...
>>
>> <https://www.youtube.com/watch?v=QP75EeFVyOI> 1st drone
>> <https://www.youtube.com/watch?v=fp5Z1Nu4Hko> 2nd drone
>> <https://www.youtube.com/watch?v=fLxE8iS7fPI> 3rd drone
>>
>> The interest with the forest is that trajectories are curved, in
>> order to avoid obstacles. It is measured thanks to a projective
>> transform, and represented with <Ry,Rz,Tx,Tz> thanks to POV-Ray.
>> The evolution of the drone is shown in front-view with its camera.
> 
> It is possible to perceive the relief (in depth) of a scene, when we
> have at least two different viewpoints of it. Here is a new example with
> a drone flying in the middle of a forest of trees, and from which we
> process the video stream from the embedded camera...
> 
> <https://www.youtube.com/watch?v=WJ20EBM3PTc>
> 
> When the two views of the same scene are distant in space, we speak
> of "spatial disparity". In the present case, the two viewpoints are
> distant in time, and we then speak of "temporal disparity". This
> involves knowing whether the two images of the same scene are acquired
> simultaneously, or delayed in time. We can perceive the relief in depth
> in this case, with a single camera and its continuous video stream.

Let remind us the starting point from this thread... We've redone the
experiment from Hernan Badino, who is walking with a camera on his head:

     <https://www.youtube.com/watch?v=GeVJMamDFXE>

Hernan determines his 2D ego-motion in the x-y plane, from corresponding
interest points that persist in the video stream. That means he is
calculating the projection matrix of the movement to deduce translations
in the ground plane. With time integration, it gives him the trajectory.

We're doing almost the same, but I work with OpenCV's optical-flow, and
not interest points. And my motion model is 3D, to obtain 8 parameters
in rotation and translation, that I can use in Persistence Of Vision.

I'm reconstituting the 3D movement, and I discover it's giving "temporal
disparity", that is depth from motion.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4924

Hi,

Here is another result...

Francois LE COAT writes:
>>>> The principle of dominant 2D motion appeared at INRIA, it is here:
>>>>
>>>> <https://www.irisa.fr/vista/Themes/Logiciel/Motion-2D/Motion-2D.html>
>>>>
>>>> In our case, the dominant motion estimated from the approximation of
>>>> optical-flow (DIS - Dense Inverse Search OpenCV) is 3D and projective.
>>>>
>>>>> The experiment from Hernan Badino was redone. You can see it there...
>>>>>
>>>>>      <https://www.youtube.com/watch?v=fqWdSfN9FiA> Source
>>>>>
>>>>> The main interest is that video is looping, and the result is almost:
>>>>>
>>>>>      <https://www.youtube.com/watch?v=0ZPJmnBh03M> Reworked
>>>>>
>>>>> Well, Hernan Badino is moving his head when he is walking, so the
>>>>> reconstructed trajectory is not perfectly looping at the end. But
>>>>> we can reconstruct the movement almost perfectly. We use OpenCV
>>>>> for image processing, and POV-Ray for 3D representation. We have
>>>>> to determine projective dominant motion in the video with a
>>>>> reference image, and change it when correlation drops below 80%.
>>>>>
>>>>> We have a 3D inertial model of motion, that's why POV-Ray helps =)
>>>
>>> Three drones are flying between forests of trees. Thanks to the
>>> optical-flow (DIS OpenCV) measured on successive images, the
>>> "temporal disparity" reveals the forest of trees (3rd dimension)...
>>>
>>> <https://www.youtube.com/watch?v=QP75EeFVyOI> 1st drone
>>> <https://www.youtube.com/watch?v=fp5Z1Nu4Hko> 2nd drone
>>> <https://www.youtube.com/watch?v=fLxE8iS7fPI> 3rd drone
>>>
>>> The interest with the forest is that trajectories are curved, in
>>> order to avoid obstacles. It is measured thanks to a projective
>>> transform, and represented with <Ry,Rz,Tx,Tz> thanks to POV-Ray.
>>> The evolution of the drone is shown in front-view with its camera.
>>
>> It is possible to perceive the relief (in depth) of a scene, when we
>> have at least two different viewpoints of it. Here is a new example with
>> a drone flying in the middle of a forest of trees, and from which we
>> process the video stream from the embedded camera...
>>
>> <https://www.youtube.com/watch?v=WJ20EBM3PTc>
>>
>> When the two views of the same scene are distant in space, we speak
>> of "spatial disparity". In the present case, the two viewpoints are
>> distant in time, and we then speak of "temporal disparity". This
>> involves knowing whether the two images of the same scene are acquired
>> simultaneously, or delayed in time. We can perceive the relief in depth
>> in this case, with a single camera and its continuous video stream.
> 
> Let remind us the starting point from this thread... We've redone the
> experiment from Hernan Badino, who is walking with a camera on his head:
> 
>      <https://www.youtube.com/watch?v=GeVJMamDFXE>
> 
> Hernan determines his 2D ego-motion in the x-y plane, from corresponding
> interest points that persist in the video stream. That means he is
> calculating the projection matrix of the movement to deduce translations
> in the ground plane. With time integration, it gives him the trajectory.
> 
> We're doing almost the same, but I work with OpenCV's optical-flow, and
> not interest points. And my motion model is 3D, to obtain 8 parameters
> in rotation and translation, that I can use in Persistence Of Vision.
> 
> I'm reconstituting the 3D movement, and I discover it's giving "temporal
> disparity", that is depth from motion.

An instrumented motorcycle rolls on the track of a speed circuit. Thanks
to the approximation of optical flow (DIS - OpenCV) by the dominant
projective movement, we determine translations on the ground plane,
roll and yaw. That is to say the trajectory by projective parameters
(Tx,Tz,Ry,Rz).

<https://www.youtube.com/watch?v=-QLJ2ke9mN8>

Image data comes from the publication:

Bastien Vincke, Pauline Michel, Abdelhafid El Ouardi, Bruno Larnaudie,
Flavien Delgehier, Rabah Sadoun, Samir Bouaziz, Stéphane Espié, Sergio
Rodriguez, Abderrahmane Boubezoul. (Dec. 2024). Real Track Experiment
Dataset for Motorcycle Rider Behavior and Trajectory Reconstruction.
Data in Brief, Vol. 57, 111026.

The instrumented motorcycle makes a complete lap of the track. The
correlation threshold is set at 90% between successive images, to
reset the calculation of the projective dynamic model.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4925

Hi,

A WEB page was made to illustrate Monocular Depth...

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

A drone flies between the trees of a forest. Thanks to the optical-flow
measured on successive images, the monocular depth reveals the
forest of trees... We take a reference image, the optical-flow is
measured on two rectified images. Then we change the reference when
the inter-correlation drops below 60%. We can perceive the relief in
depth with a single camera, over time.

In fact, when we watch images captured by a drone, although there is
only one camera, we often see the relief. This is particularly marked
for trees in a forest. The goal here is to evaluate this relief, with
a measurement of "optical-flow", which allows one image to be matched
with another, when they seem to be close (we say they are "correlated").

We have two eyes, and the methods for measuring visible relief by
stereoscopy are very developed. Since the beginning of photography,
there were devices like the “stereoscope” which allows you to see the
relief with two pictures, naturally. It is possible to measure relief,
thanks to epipolar geometry, and well-known mathematics. There are many
measurement methods, very effective and based on human vision.

When it comes to measuring relief with a single camera, knowledge is
less established. There are 3D cameras, called "RGBD" with a "D" for
"depth". But how do they work? Is it possible to improve those? What
we are showing here does not require the use of any “artificial neural
network”. It is a physical measurement, with a classic algorithm,
which does not come from A.I. nor a big computer :-)

This is about measuring monocular depth, just as stereoscopic disparity
is measured. It means quantifying the depth, with images from a single
camera. We can see this relief naturally, but it is a matter of
measuring it with the optical-flow.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4926

Hi,

Francois LE COAT writes:
> A WEB page was made to illustrate Monocular Depth...
> 
> <https://hebergement.universite-paris-saclay.fr/lecoat/demoweb/monocular_depth.html> 
> 
> 
> A drone flies between the trees of a forest. Thanks to the optical-flow
> measured on successive images, the monocular depth reveals the
> forest of trees... We take a reference image, the optical-flow is
> measured on two rectified images. Then we change the reference when
> the inter-correlation drops below 60%. We can perceive the relief in
> depth with a single camera, over time.
> 
> In fact, when we watch images captured by a drone, although there is
> only one camera, we often see the relief. This is particularly marked
> for trees in a forest. The goal here is to evaluate this relief, with
> a measurement of "optical-flow", which allows one image to be matched
> with another, when they seem to be close (we say they are "correlated").
> 
> We have two eyes, and the methods for measuring visible relief by
> stereoscopy are very developed. Since the beginning of photography,
> there were devices like the “stereoscope” which allows you to see the
> relief with two pictures, naturally. It is possible to measure relief,
> thanks to epipolar geometry, and well-known mathematics. There are many
> measurement methods, very effective and based on human vision.
> 
> When it comes to measuring relief with a single camera, knowledge is
> less established. There are 3D cameras, called "RGBD" with a "D" for
> "depth". But how do they work? Is it possible to improve those? What
> we are showing here does not require the use of any “artificial neural
> network”. It is a physical measurement, with a classic algorithm,
> which does not come from A.I. nor a big computer :-)
> 
> This is about measuring monocular depth, just as stereoscopic disparity
> is measured. It means quantifying the depth, with images from a single
> camera. We can see this relief naturally, but it is a matter of
> measuring it with the optical-flow.

Until now, drone images came from forests in France. The first images
were obtained in the French Vosges.

<https://www.youtube.com/watch?v=245yJJrwMQ0> Drone in the forest

We are now seeing more and more drones in forests outside of France.
The available image sources are diversifying...

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4927

Hi,

Francois LE COAT writes:
> Until now, drone images came from forests in France. The first images
> were obtained in the French Vosges.
> 
> <https://www.youtube.com/watch?v=245yJJrwMQ0> Drone in the forest
> 
> We are now seeing more and more drones in forests outside of France.
> The available image sources are diversifying...

Here is a sequence of images of a ballad in the forest. This scene
is observed by a tracking drone...

     <https://www.youtube.com/watch?v=46VWJ6-YqtY>

The camera's movement is estimated in the images using a projective
dominant motion measurement. The presence of a man in the image sequence
does not interfere with trajectory estimation, because the character
occupies a position in the field of vision that is not dominant. The
dominant motion corresponds to the scrolling of the scenery, that is
the movement of the forest relative to an observing camera.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4928

Hi,

Francois LE COAT writes:
>> Until now, drone images came from forests in France. The first images
>> were obtained in the French Vosges.
>>
>> <https://www.youtube.com/watch?v=245yJJrwMQ0> Drone in the forest
>>
>> We are now seeing more and more drones in forests outside of France.
>> The available image sources are diversifying...
> 
> Here is a sequence of images of a ballad in the forest. This scene
> is observed by a tracking drone...
> 
>      <https://www.youtube.com/watch?v=46VWJ6-YqtY>
> 
> The camera's movement is estimated in the images using a projective
> dominant motion measurement. The presence of a man in the image sequence
> does not interfere with trajectory estimation, because the character
> occupies a position in the field of vision that is not dominant. The
> dominant motion corresponds to the scrolling of the scenery, that is
> the movement of the forest relative to an observing camera.

Here is a drone in the forest...

<https://www.youtube.com/watch?v=h3vhlRBB9tg> Forest

We also obtain the trajectory in space: <https://skfb.ly/pxGqL>

It's interesting to note that this trajectory loops. That is to say,
the drone passes over the professional pilot's location, and is in
the same place at the end of the video.

This proves the quality of the trajectory estimation in space. :-)

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4929

Hi,

Francois LE COAT writes:
>>> Until now, drone images came from forests in France. The first images
>>> were obtained in the French Vosges.
>>>
>>> <https://www.youtube.com/watch?v=245yJJrwMQ0> Drone in the forest
>>>
>>> We are now seeing more and more drones in forests outside of France.
>>> The available image sources are diversifying...
>>
>> Here is a sequence of images of a ballad in the forest. This scene
>> is observed by a tracking drone...
>>
>>      <https://www.youtube.com/watch?v=46VWJ6-YqtY>
>>
>> The camera's movement is estimated in the images using a projective
>> dominant motion measurement. The presence of a man in the image sequence
>> does not interfere with trajectory estimation, because the character
>> occupies a position in the field of vision that is not dominant. The
>> dominant motion corresponds to the scrolling of the scenery, that is
>> the movement of the forest relative to an observing camera.
> 
> Here is a drone in the forest...
> 
> <https://www.youtube.com/watch?v=h3vhlRBB9tg> Forest
> 
> We also obtain the trajectory in space: <https://skfb.ly/pxGqL>
> 
> It's interesting to note that this trajectory loops. That is to say,
> the drone passes over the professional pilot's location, and is in
> the same place at the end of the video.
> 
> This proves the quality of the trajectory estimation in space. :-)

Here's a long drone flight through a Swedish forest...

<https://www.youtube.com/watch?v=ppW5BbDPFHc> Swedish forest

We obtain the estimated trajectory in space: <https://skfb.ly/pxZHy>

The image matching algorithm doesn't incorporate any prior knowledge
about what the camera is observing. This might look like a SLAM
(Simultaneous Localization And Mapping) method, which is a sparse
method, but what is presented is a global and dense method based
on optical-flow measurement (Dense Inverse Search - DIS).

This is an algorithm, i.e. a numerical recipe, that does not use
artificial neural networks. We obtain a filtered measurement (by
a Kalman filter) of physical data.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4930

Hi,

Here's a drone's long flight through the forest in France...

	<https://www.youtube.com/watch?v=f_Zi811qWFI>

We're in rather complicated lighting conditions, with shadows and
clouds. Hence the significant noise in the video. The optical-flow
(DIS - OpenCV) that allows the determination of monocular depth
performs rather well. Mathematicians refer to this determination
as an "ill-posed problem". But statistically, for the large images
we're dealing with, it works well :-)

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4931

Hi,

Francois LE COAT wrote:
> Here's a drone's long flight through the forest in France...
> 
>      <https://www.youtube.com/watch?v=f_Zi811qWFI>
> 
> We're in rather complicated lighting conditions, with shadows and
> clouds. Hence the significant noise in the video. The optical-flow
> (DIS - OpenCV) that allows the determination of monocular depth
> performs rather well. Mathematicians refer to this determination
> as an "ill-posed problem". But statistically, for the large images
> we're dealing with, it works well :-)

Here's a sequence of images from a drone in the forest.

	<https://www.youtube.com/watch?v=gOAwt0I7GNE>

These image computing looks like SLAM (Simultaneous Localization and
Mapping) method. We obtain both the location (trajectory) and the
visible relief (3D depth map). However, we're dealing with a monocular
(single camera) image sequence, not a stereoscopic (human vision) one.

Here's the drone's trajectory in space: <https://skfb.ly/pCyBy>

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4932

Hi,

Francois LE COAT writes:
>> Here's a drone's long flight through the forest in France...
>>
>>      <https://www.youtube.com/watch?v=f_Zi811qWFI>
>>
>> We're in rather complicated lighting conditions, with shadows and
>> clouds. Hence the significant noise in the video. The optical-flow
>> (DIS - OpenCV) that allows the determination of monocular depth
>> performs rather well. Mathematicians refer to this determination
>> as an "ill-posed problem". But statistically, for the large images
>> we're dealing with, it works well :-)
> 
> Here's a sequence of images from a drone in the forest.
> 
>      <https://www.youtube.com/watch?v=gOAwt0I7GNE>
> 
> These image computing looks like SLAM (Simultaneous Localization and
> Mapping) method. We obtain both the location (trajectory) and the
> visible relief (3D depth map). However, we're dealing with a monocular
> (single camera) image sequence, not a stereoscopic (human vision) one.
> 
> Here's the drone's trajectory in space: <https://skfb.ly/pCyBy>

Here's a sequence of images from a drone in the forest:

	<https://www.youtube.com/watch?v=ToRk-o1cD_4>

This image computing looks like SLAM (Simultaneous Localization and
Mapping) method. We obtain both the location (trajectory) and the
visible relief (3D depth map). However, we're dealing with a monocular
(single camera) image sequence, not a stereoscopic (human vision) one.

The question that arises, then, is why do highly accurate inertial
navigation systems drift? Why are loop-closing methods, used with the
SLAM algorithm? Would an optical inertial navigation system be subject
to drift in the trajectory it estimates? That's debatable, isn't it?

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4933

Hi,

This is a large sequence of images from a drone in the forest:

<https://www.youtube.com/watch?v=22QoxEwMBQA> The forest

This image computing looks like SLAM (Simultaneous Localization and
Mapping) method. We obtain both the location (trajectory) and the
visible relief (3D depth map). However, we're dealing with a monocular
(single camera) image sequence, not a stereoscopic (human vision) one.

The sequence is constituted of 10 000 images at 60 frames per second.
It is both large and high-resolution.

Here's the drone's trajectory in space: <https://skfb.ly/pDCwR>

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4934

Hi,

Francois LE COAT writes:
> This is a large sequence of images from a drone in the forest:
> 
> <https://www.youtube.com/watch?v=22QoxEwMBQA> The forest
> 
> This image computing looks like SLAM (Simultaneous Localization and
> Mapping) method. We obtain both the location (trajectory) and the
> visible relief (3D depth map). However, we're dealing with a monocular
> (single camera) image sequence, not a stereoscopic (human vision) one.
> 
> The sequence is constituted of 10 000 images at 60 frames per second.
> It is both large and high-resolution.
> 
> Here's the drone's trajectory in space: <https://skfb.ly/pDCwR>

This is a sequence of images from a drone in the forest:

<https://www.youtube.com/watch?v=Kt0mkFYX45A> Drone

The drone flies very slowly compared to the acquisition rate of 30
frames per second. This makes it much easier to observe the trajectory
and visible depth. The successive images in the sequence are highly
correlated. It is not necessary to frequently reset the reference image
selection over time, as the correlation between this image and
subsequent images remains well above the 60% threshold chosen for
resetting.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4935

Hi,

Francois LE COAT writes:
> This is a sequence of images from a drone in the forest:
> 
> <https://www.youtube.com/watch?v=Kt0mkFYX45A> Drone
> 
> The drone flies very slowly compared to the acquisition rate of 30
> frames per second. This makes it much easier to observe the trajectory
> and visible depth. The successive images in the sequence are highly
> correlated. It is not necessary to frequently reset the reference image
> selection over time, as the correlation between this image and
> subsequent images remains well above the 60% threshold chosen for
> resetting.

Here's a sequence of images from a drone in the forest:

<https://www.youtube.com/watch?v=hFcDUU626po> Swedish woods

This image computing called 3D Optical Inertia looks like SLAM
(Simultaneous Localization and Mapping) method. We obtain both
the location (trajectory) and the visible relief (3D depth map).
However, we're dealing with a monocular (single camera) image
sequence, not a stereoscopic (human vision) one.

Here is the spline trajectory in space: <https://skfb.ly/pFx6u>

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4936

Hi,

Here is a sequence of images of a ballad in the forest. This scene is
observed by a tracking drone...

	<https://www.youtube.com/watch?v=b4aTYv7A1lM>

The camera's movement is estimated in the images using a projective
dominant motion measurement. The presence of a man in the image sequence
does not interfere with trajectory estimation, because the character
occupies a position in the field of vision that is not dominant. The
dominant motion corresponds to the scrolling of the scenery, that is
the movement of an observing camera relative to the forest.

Best regards,

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

[toc] | [prev] | [next] | [standalone]

#4937

Hi,

Here is the flight of a drone through the forest...

	<https://www.youtube.com/watch?v=-RefA0o2wkE>

What is "temporal disparity"?

In the context of stereoscopic vision or image matching, disparity
(typically spatial) measures the difference in position of a single
point across two images (e.g., left and right).

When applied to the temporal domain, temporal disparity therefore
refers to:
- The positional difference of a point or feature between two
consecutive images in a video sequence, reflecting motion or
temporal change.
- A metric used in computer vision to estimate optical-flow or
3D structure from image sequences.

Measuring optical-flow between successive images matched over time
— by estimating projective dominant motion — reveals "temporal
disparity".

Best regards,

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

[toc] | [prev] | [standalone]

Back to top | Article view | sci.image.processing

csiph-web