rendering a 3D image from side views only

classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

rendering a 3D image from side views only

BananaBoy
Hi, I'm pretty new to imageJ and image analysis in general but I have a
question that I cannot seem to find a clear answer to which is the
following:

I would like to construct a 3D model of a plant's root system by taking side
views (and potentially a top view) images. For obvious reasons, I cannot
create a stack of slices along the vertical axis. Is this possible?

For example, I could create a setup where the plant is at a fixed position
from the camera (with naked and clean roots hanging in the air). I could
then rotate the plant 360 degrees in increments of 10 degrees, so 35
pictures in total.

Is ImageJ capable of rendering a 3D model of this root system?


Thank you.



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Herbie
Good day!

This post was cross-posted to the IJ-Forum:
<https://forum.image.sc/t/3d-model-without-stacks/30598>

If you think of tomographic reconstruction you must specify the maximum
number N of  linear independent resolution elements across the diameter
of your top view. Given this number, the minimum number of angle
increments z that is necessary for the adequate reconstruction of
z-slices is:

z < N * π * 0.5

I don't think that 360 / z = 10deg will be sufficient.

Regards

Herbie

:::::::::::::::::::::::::::::::::::::::
Am 21.10.19 um 22:38 schrieb BananaBoy:

> Hi, I'm pretty new to imageJ and image analysis in general but I have a
> question that I cannot seem to find a clear answer to which is the
> following:
>
> I would like to construct a 3D model of a plant's root system by taking side
> views (and potentially a top view) images. For obvious reasons, I cannot
> create a stack of slices along the vertical axis. Is this possible?
>
> For example, I could create a setup where the plant is at a fixed position
> from the camera (with naked and clean roots hanging in the air). I could
> then rotate the plant 360 degrees in increments of 10 degrees, so 35
> pictures in total.
>
> Is ImageJ capable of rendering a 3D model of this root system?
>
>
> Thank you.
>
>
>
> --
> Sent from: http://imagej.1557.x6.nabble.com/
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Herbie
Sorry it must read:

z > N * π * 0.5

Regards

Herbie

::::::::::::::::::::::::::::::::::::
Am 21.10.19 um 23:29 schrieb Herbie:

> Good day!
>
> This post was cross-posted to the IJ-Forum:
> <https://forum.image.sc/t/3d-model-without-stacks/30598>
>
> If you think of tomographic reconstruction you must specify the maximum
> number N of  linear independent resolution elements across the diameter
> of your top view. Given this number, the minimum number of angle
> increments z that is necessary for the adequate reconstruction of
> z-slices is:
>
> z > N * π * 0.5
>
> I don't think that 360 / z = 10deg will be sufficient.
>
> Regards
>
> Herbie
>
> :::::::::::::::::::::::::::::::::::::::
> Am 21.10.19 um 22:38 schrieb BananaBoy:
>> Hi, I'm pretty new to imageJ and image analysis in general but I have a
>> question that I cannot seem to find a clear answer to which is the
>> following:
>>
>> I would like to construct a 3D model of a plant's root system by
>> taking side
>> views (and potentially a top view) images. For obvious reasons, I cannot
>> create a stack of slices along the vertical axis. Is this possible?
>>
>> For example, I could create a setup where the plant is at a fixed
>> position
>> from the camera (with naked and clean roots hanging in the air). I could
>> then rotate the plant 360 degrees in increments of 10 degrees, so 35
>> pictures in total.
>>
>> Is ImageJ capable of rendering a 3D model of this root system?
>>
>>
>> Thank you.
>>
>>
>>
>> --
>> Sent from: http://imagej.1557.x6.nabble.com/
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Herbie
Good day!

As a follow-up regarding "tomographic reconstruction", i.e. "image
reconstruction from projections", proposed on the IJ-Forum, I should
like to mention that "tomographic reconstruction" is *not* applicable in
the described case because there are no projections but only side views
of the visible (reflected light) structures. From the latter it is
impossible to obtain "tomographic reconstruction".

Regards

Herbie

::::::::::::::::::::::::::::::::::::
Am 21.10.19 um 23:47 schrieb Herbie:

> Sorry it must read:
>
> z > N * π * 0.5
>
> Regards
>
> Herbie
>
> ::::::::::::::::::::::::::::::::::::
> Am 21.10.19 um 23:29 schrieb Herbie:
>> Good day!
>>
>> This post was cross-posted to the IJ-Forum:
>> <https://forum.image.sc/t/3d-model-without-stacks/30598>
>>
>> If you think of tomographic reconstruction you must specify the
>> maximum number N of  linear independent resolution elements across the
>> diameter of your top view. Given this number, the minimum number of
>> angle increments z that is necessary for the adequate reconstruction
>> of z-slices is:
>>
>> z > N * π * 0.5
>>
>> I don't think that 360 / z = 10deg will be sufficient.
>>
>> Regards
>>
>> Herbie
>>
>> :::::::::::::::::::::::::::::::::::::::
>> Am 21.10.19 um 22:38 schrieb BananaBoy:
>>> Hi, I'm pretty new to imageJ and image analysis in general but I have a
>>> question that I cannot seem to find a clear answer to which is the
>>> following:
>>>
>>> I would like to construct a 3D model of a plant's root system by
>>> taking side
>>> views (and potentially a top view) images. For obvious reasons, I cannot
>>> create a stack of slices along the vertical axis. Is this possible?
>>>
>>> For example, I could create a setup where the plant is at a fixed
>>> position
>>> from the camera (with naked and clean roots hanging in the air). I could
>>> then rotate the plant 360 degrees in increments of 10 degrees, so 35
>>> pictures in total.
>>>
>>> Is ImageJ capable of rendering a 3D model of this root system?
>>>
>>>
>>> Thank you.
>>>
>>>
>>>
>>> --
>>> Sent from: http://imagej.1557.x6.nabble.com/
>>>
>>> --
>>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>>
>>
>> --
>> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>>
>
> --
> ImageJ mailing list: http://imagej.nih.gov/ij/list.html
>

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Kenneth R Sloan-2
a) No - ImageJ does not appear to directly support 3D reconstruction from multiple side views.

b) If you want to implement it yourself, I would encourage you to start at:

        https://en.wikipedia.org/wiki/3D_reconstruction_from_multiple_images <https://en.wikipedia.org/wiki/3D_reconstruction_from_multiple_images>

A lot will depend on how much detail you require, and how much annotation you can provide for the images

--
Kenneth Sloan
[hidden email]
Vision is the art of seeing what is invisible to others.






--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Dr. Michael DOUBE
In reply to this post by BananaBoy
Hi Bill,

What you want is photogrammetry, which is the process of generating surface models (2.5D if you like) from a series of photographs. It's a popular technique in museum digital curation because it's no destructive and the hardware requirements are minimal. In my current department, a whole Omura's whale skeleton was digitised with just a digital SLR camera and some software (I don't know which one, sorry). You are likely to need many more photographs than you think, because you have to gather photons from every surface, otherwise surface regions will be missing from the final mesh. There are a few libraries out there that might be helpful for you - not sure if any are implemented in ImageJ.

As others have pointed out, tomography requires images projected through the complete thickness of the sample for reconstruction at many angles - usually 1 - 3 thousand per full rotation, but you can get away with fewer with special reconstruction algorithms. There is a group in the UK, at the Hounsfield lab, doing X-ray microtomogrpahy of root systems. Would be worth checking them out.

Hope it helps,

Michael

On 22/10/2019 04:38, BananaBoy wrote:

Hi, I'm pretty new to imageJ and image analysis in general but I have a
question that I cannot seem to find a clear answer to which is the
following:

I would like to construct a 3D model of a plant's root system by taking side
views (and potentially a top view) images. For obvious reasons, I cannot
create a stack of slices along the vertical axis. Is this possible?

For example, I could create a setup where the plant is at a fixed position
from the camera (with naked and clean roots hanging in the air). I could
then rotate the plant 360 degrees in increments of 10 degrees, so 35
pictures in total.

Is ImageJ capable of rendering a 3D model of this root system?


Thank you.



--
Sent from: http://imagej.1557.x6.nabble.com/

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html


--
[Jockey Club College of Veterinary Medicine and                      Life Sciences - City University of Hong Kong]



Dr. Michael DOUBE
BVSc, BPhil, PhD, PGCert Vet Ed, MRCVS, FRMS

Associate Professor
Department of Infectious Diseases
and Public Health

5/F, Block 2, To Yuen Building
31 To Yuen Street
Kowloon, Hong Kong
Tel: (852) 3442 5296
Fax: (852) 3442 0589
Website: www.cityu.edu.hk/ph<http://www.cityu.edu.hk/ph>



Disclaimer: This email (including any attachments) is for the use of the intended recipient only and may contain confidential information and/or copyright material. If you are not the intended recipient, please notify the sender immediately and delete this email and all copies from your system. Any unauthorized use, disclosure, reproduction, copying, distribution, or other form of unauthorized dissemination of the contents is expressly prohibited.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Stein Rørvik
Michael is right that in what you need is photogrammetry, not tomography. The tomography requires that your images are projections; that is, an image that represents some kind of sum of density values along a path, like an X-Ray image. That will not be the case for a "normal" photographic image.

I have done a lot of X-ray microtomography but also some photogrammetry of various objects. For photogrammetry, I have used a commercial software called Agisoft PhotoScan that is easy to use and give excellent results. You need something like 500 images to get a reasonably good dataset, probably more in your case. So you need a rig with automated rotation and image acquisition for reproducible results. It is also probably possible to find free / open source software for doing this, but I have not checked that out since the Agisoft software works so well. If you only have a few samples and do not want to invest your time in setting up a rig, you can take thousands of images in less than an hour with a normal SLR and a big memory card. Just walk around your object and snap random photos at as many angles as you have the patience to do. Then feed it into the software and the rest is automatic.

A disadvantage with Agisoft PhotoScan is that it requires your scene to be fixed and the camera moving. So it will NOT work well with a setup using a fixed camera and rotating object. The reason is that it uses details in the surroundings of the object to calculate the orientation of the camera towards the object. That might have changed in recent updates though; I have not checked recently.

Stein

-----Original Message-----
From: ImageJ Interest Group <[hidden email]> On Behalf Of Dr. Michael DOUBE
Sent: 23. oktober 2019 02:50
To: [hidden email]
Subject: Re: rendering a 3D image from side views only

Hi Bill,

What you want is photogrammetry, which is the process of generating surface models (2.5D if you like) from a series of photographs. It's a popular technique in museum digital curation because it's no destructive and the hardware requirements are minimal. In my current department, a whole Omura's whale skeleton was digitised with just a digital SLR camera and some software (I don't know which one, sorry). You are likely to need many more photographs than you think, because you have to gather photons from every surface, otherwise surface regions will be missing from the final mesh. There are a few libraries out there that might be helpful for you - not sure if any are implemented in ImageJ.

As others have pointed out, tomography requires images projected through the complete thickness of the sample for reconstruction at many angles - usually 1 - 3 thousand per full rotation, but you can get away with fewer with special reconstruction algorithms. There is a group in the UK, at the Hounsfield lab, doing X-ray microtomogrpahy of root systems. Would be worth checking them out.

Hope it helps,

Michael

On 22/10/2019 04:38, BananaBoy wrote:

Hi, I'm pretty new to imageJ and image analysis in general but I have a question that I cannot seem to find a clear answer to which is the
following:

I would like to construct a 3D model of a plant's root system by taking side views (and potentially a top view) images. For obvious reasons, I cannot create a stack of slices along the vertical axis. Is this possible?

For example, I could create a setup where the plant is at a fixed position from the camera (with naked and clean roots hanging in the air). I could then rotate the plant 360 degrees in increments of 10 degrees, so 35 pictures in total.

Is ImageJ capable of rendering a 3D model of this root system?


Thank you.



--
Sent from: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C71400840f30a4da2a4ac08d757532465%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637073886924693790&amp;sdata=wQfEHq7TDA3KHgMq29EOW1uOjLxVr1D7dB4U8poHMA0%3D&amp;reserved=0

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C71400840f30a4da2a4ac08d757532465%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637073886924693790&amp;sdata=eNKYuKmWgUnstfu9r8kJ4TZFUvmPyLn0vRCNOxf5wJY%3D&amp;reserved=0


--
[Jockey Club College of Veterinary Medicine and                      Life Sciences - City University of Hong Kong]



Dr. Michael DOUBE
BVSc, BPhil, PhD, PGCert Vet Ed, MRCVS, FRMS

Associate Professor
Department of Infectious Diseases
and Public Health

5/F, Block 2, To Yuen Building
31 To Yuen Street
Kowloon, Hong Kong
Tel: (852) 3442 5296
Fax: (852) 3442 0589
Website: https://eur03.safelinks.protection.outlook.com/?url=www.cityu.edu.hk%2Fph&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C71400840f30a4da2a4ac08d757532465%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637073886924703785&amp;sdata=DfQcBCGtw5BXcGTU7t6h2qKqk384n2fv6Cq6wkt%2Baus%3D&amp;reserved=0<https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.cityu.edu.hk%2Fph&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C71400840f30a4da2a4ac08d757532465%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637073886924703785&amp;sdata=5zpUfQrAb5cTvA%2BP0FKAIHg8CKoDKg8xV%2FL7rLUD%2Bds%3D&amp;reserved=0>



Disclaimer: This email (including any attachments) is for the use of the intended recipient only and may contain confidential information and/or copyright material. If you are not the intended recipient, please notify the sender immediately and delete this email and all copies from your system. Any unauthorized use, disclosure, reproduction, copying, distribution, or other form of unauthorized dissemination of the contents is expressly prohibited.

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C71400840f30a4da2a4ac08d757532465%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C1%7C637073886924703785&amp;sdata=%2FzQeXsxFti9uK%2F5wgN3hqw%2FA0KoSJKgxKmnB%2BTFuzic%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html
Reply | Threaded
Open this post in threaded view
|

Re: rendering a 3D image from side views only

Stein Rørvik
A comment to my own post; I looked on the Agisoft website and they are explaining how to overcome the problem with using a fixed camera and rotating object in a tutorial explaining some background masking functionality, which was introduced recently. However, this might be difficult to do with such a highly detailed object a plant root is.

You might therefore be better off with a tomography based solution, depending on your budget. One tomography scan might cost the same as one SLR camera, or twice the cost of a license for the photogrammetry software. A great advantage with tomography data though is that it can be processed directly in ImageJ, while the photogrammetry software will produce a point cloud or surface model which  needs to be converted.

Stein

-----Original Message-----
From: ImageJ Interest Group <[hidden email]> On Behalf Of Stein Rørvik
Sent: 24. oktober 2019 16:10
To: [hidden email]
Subject: Re: rendering a 3D image from side views only

Michael is right that in what you need is photogrammetry, not tomography. The tomography requires that your images are projections; that is, an image that represents some kind of sum of density values along a path, like an X-Ray image. That will not be the case for a "normal" photographic image.

I have done a lot of X-ray microtomography but also some photogrammetry of various objects. For photogrammetry, I have used a commercial software called Agisoft PhotoScan that is easy to use and give excellent results. You need something like 500 images to get a reasonably good dataset, probably more in your case. So you need a rig with automated rotation and image acquisition for reproducible results. It is also probably possible to find free / open source software for doing this, but I have not checked that out since the Agisoft software works so well. If you only have a few samples and do not want to invest your time in setting up a rig, you can take thousands of images in less than an hour with a normal SLR and a big memory card. Just walk around your object and snap random photos at as many angles as you have the patience to do. Then feed it into the software and the rest is automatic.

A disadvantage with Agisoft PhotoScan is that it requires your scene to be fixed and the camera moving. So it will NOT work well with a setup using a fixed camera and rotating object. The reason is that it uses details in the surroundings of the object to calculate the orientation of the camera towards the object. That might have changed in recent updates though; I have not checked recently.

Stein

-----Original Message-----
From: ImageJ Interest Group <[hidden email]> On Behalf Of Dr. Michael DOUBE
Sent: 23. oktober 2019 02:50
To: [hidden email]
Subject: Re: rendering a 3D image from side views only

Hi Bill,

What you want is photogrammetry, which is the process of generating surface models (2.5D if you like) from a series of photographs. It's a popular technique in museum digital curation because it's no destructive and the hardware requirements are minimal. In my current department, a whole Omura's whale skeleton was digitised with just a digital SLR camera and some software (I don't know which one, sorry). You are likely to need many more photographs than you think, because you have to gather photons from every surface, otherwise surface regions will be missing from the final mesh. There are a few libraries out there that might be helpful for you - not sure if any are implemented in ImageJ.

As others have pointed out, tomography requires images projected through the complete thickness of the sample for reconstruction at many angles - usually 1 - 3 thousand per full rotation, but you can get away with fewer with special reconstruction algorithms. There is a group in the UK, at the Hounsfield lab, doing X-ray microtomogrpahy of root systems. Would be worth checking them out.

Hope it helps,

Michael

On 22/10/2019 04:38, BananaBoy wrote:

Hi, I'm pretty new to imageJ and image analysis in general but I have a question that I cannot seem to find a clear answer to which is the
following:

I would like to construct a 3D model of a plant's root system by taking side views (and potentially a top view) images. For obvious reasons, I cannot create a stack of slices along the vertical axis. Is this possible?

For example, I could create a setup where the plant is at a fixed position from the camera (with naked and clean roots hanging in the air). I could then rotate the plant 360 degrees in increments of 10 degrees, so 35 pictures in total.

Is ImageJ capable of rendering a 3D model of this root system?


Thank you.



--
Sent from: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.1557.x6.nabble.com%2F&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C38037788d1fe423efe9608d7588c3709%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637075231558881233&amp;sdata=4bJogvda3o6JHjgKPzQABfG%2F0w0viVt%2BRmXlPfZadYg%3D&amp;reserved=0

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C38037788d1fe423efe9608d7588c3709%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637075231558881233&amp;sdata=sjONCC4dtXthGqizvdaA%2F6y8y4lw%2BxfCp8jBUNy60Sc%3D&amp;reserved=0


--
[Jockey Club College of Veterinary Medicine and                      Life Sciences - City University of Hong Kong]



Dr. Michael DOUBE
BVSc, BPhil, PhD, PGCert Vet Ed, MRCVS, FRMS

Associate Professor
Department of Infectious Diseases
and Public Health

5/F, Block 2, To Yuen Building
31 To Yuen Street
Kowloon, Hong Kong
Tel: (852) 3442 5296
Fax: (852) 3442 0589
Website: https://eur03.safelinks.protection.outlook.com/?url=www.cityu.edu.hk%2Fph&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C38037788d1fe423efe9608d7588c3709%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637075231558881233&amp;sdata=YLRcvGfigz7MrKvhnLYkxNTomyrIIncyeqKceqLdlf4%3D&amp;reserved=0<https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.cityu.edu.hk%2Fph&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C38037788d1fe423efe9608d7588c3709%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637075231558881233&amp;sdata=7G6p%2By%2BcY7sq4n2vREQFP1JFajfGdbsFb56wv%2BRBOYU%3D&amp;reserved=0>



Disclaimer: This email (including any attachments) is for the use of the intended recipient only and may contain confidential information and/or copyright material. If you are not the intended recipient, please notify the sender immediately and delete this email and all copies from your system. Any unauthorized use, disclosure, reproduction, copying, distribution, or other form of unauthorized dissemination of the contents is expressly prohibited.

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C38037788d1fe423efe9608d7588c3709%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637075231558881233&amp;sdata=sjONCC4dtXthGqizvdaA%2F6y8y4lw%2BxfCp8jBUNy60Sc%3D&amp;reserved=0

--
ImageJ mailing list: https://eur03.safelinks.protection.outlook.com/?url=http%3A%2F%2Fimagej.nih.gov%2Fij%2Flist.html&amp;data=02%7C01%7Cstein.rorvik%40sintef.no%7C38037788d1fe423efe9608d7588c3709%7Ce1f00f39604145b0b309e0210d8b32af%7C1%7C0%7C637075231558881233&amp;sdata=sjONCC4dtXthGqizvdaA%2F6y8y4lw%2BxfCp8jBUNy60Sc%3D&amp;reserved=0

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html