Opticks project in Gsoc 2013

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

Opticks project in Gsoc 2013

vishal tiwari
Hi!
I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

This is what i have understood about the spectral signature tool:
1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.
2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.
3. These two types can further have many subdivisions such as:
a) Graphical: 
It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.
Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.
(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)
b)Tabular: This will contain a matrix showing the value of one of the following:
Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

i wanted to ask, does the project also contain the part, for creating signatures?  

Please tell me if i am thinking in the right direction.

Thanks and kind regards,

-- 
Vishal Tiwari
Undergraduate
Computer Science and Engineering
Lab for Spatial Informatics
IIIT-Hyderabad

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

dadkins
Administrator

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad


This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

vishal tiwari
Hi!
Thanks for replying.
I downloaded the spectral plugin, and saw how the signatures can be created, from an AOI or just using individual pixels. I was thinking a seed pixel can be used to select a number of pixels in its neighborhood by using one of the pixel neighborhood function(plane 4 neighbor or diagonal 8 neighbor), to create a signature. What do you think about it?

The dendogram, is just a tree showing the clustering results. In the spectral evaluation, we want to show how the clusters and divided/grouped until we have a desired number of cluster groups(more clusters or less clusters depending if clusters are grouped or divided). I think hierarchical clustering of signatures is required here. And a bottom-up approach seems more reasonable here. I want to know more about the clustering algorithm that was implemented last year.

Thanks and kind regards,



On 25 March 2013 16:38, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs




--
Vishal Tiwari
Undergraduate
Computer Science and Engineering
Lab for Spatial Informatics
IIIT-Hyderabad

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

dadkins
Administrator

An AOI seed pixel is not nearly interesting enough for an entire GSoC project as it would take less than a week to implement.

I probably would not bother to implement it as the user can simply use the rectangle draw tool to do what you are suggesting.

 

 

As for the clustering, you will want to check out two of last year’s GSoC students, particularly the first one:

http://opticks.org/confluence/display/~himsin/Home

http://opticks.org/confluence/display/~mohitkharb/Home

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Tuesday, March 26, 2013 7:40 AM
To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

Thanks for replying.

I downloaded the spectral plugin, and saw how the signatures can be created, from an AOI or just using individual pixels. I was thinking a seed pixel can be used to select a number of pixels in its neighborhood by using one of the pixel neighborhood function(plane 4 neighbor or diagonal 8 neighbor), to create a signature. What do you think about it?

 

The dendogram, is just a tree showing the clustering results. In the spectral evaluation, we want to show how the clusters and divided/grouped until we have a desired number of cluster groups(more clusters or less clusters depending if clusters are grouped or divided). I think hierarchical clustering of signatures is required here. And a bottom-up approach seems more reasonable here. I want to know more about the clustering algorithm that was implemented last year.

 

Thanks and kind regards,

 

 

On 25 March 2013 16:38, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

 
This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad


This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

vishal tiwari
i wasn't saying making a AOI seed pixel for an entire summer :P. It might be integrated with the current methods of selecting an AOI. Today when i was using the spectral processing extension, it can to my mind. 

thanks for the references.

Thanks and kind regards,  


On 27 March 2013 00:59, Adkins, Dustan <[hidden email]> wrote:

An AOI seed pixel is not nearly interesting enough for an entire GSoC project as it would take less than a week to implement.

I probably would not bother to implement it as the user can simply use the rectangle draw tool to do what you are suggesting.

 

 

As for the clustering, you will want to check out two of last year’s GSoC students, particularly the first one:

http://opticks.org/confluence/display/~himsin/Home

http://opticks.org/confluence/display/~mohitkharb/Home

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Tuesday, March 26, 2013 7:40 AM
To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

Thanks for replying.

I downloaded the spectral plugin, and saw how the signatures can be created, from an AOI or just using individual pixels. I was thinking a seed pixel can be used to select a number of pixels in its neighborhood by using one of the pixel neighborhood function(plane 4 neighbor or diagonal 8 neighbor), to create a signature. What do you think about it?

 

The dendogram, is just a tree showing the clustering results. In the spectral evaluation, we want to show how the clusters and divided/grouped until we have a desired number of cluster groups(more clusters or less clusters depending if clusters are grouped or divided). I think hierarchical clustering of signatures is required here. And a bottom-up approach seems more reasonable here. I want to know more about the clustering algorithm that was implemented last year.

 

Thanks and kind regards,

 

 

On 25 March 2013 16:38, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

 
This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs




--
Vishal Tiwari
Undergraduate
Computer Science and Engineering
Lab for Spatial Informatics
IIIT-Hyderabad

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

vishal tiwari
Hi!
i went through the ISODATA clustering algorithm, and the tutorial of its plugin at: http://opticks.org/confluence/display/~himsin/ISODATA+Plugin+Tutorial .

Here is a my idea for making the dendrograms: 
The initial number of clusters will be set to one. And as the algorithm proceeds, the number of clusters will increase. Now in each iteration, whenever the cluster splits, the dendrogram also splits, representing the two different set of clusters as the branches of dendrogram. The min standard deviation for split will be set to the max(standard deviation of the signatures) so that , once we have the signatures, it doesn't split further. It will be a top-bottom dendrogram. And at each level of the dendrogram at each branching we can have a node which will show the percentage/ratio of the count of a signatures pixel at that level. for example, if i have two signatures A and B, then at the top level, when there is no split(just one cluster) then it would display 100% A and 100% B, after an iteration, when the cluster splits in two clusters, each branch node will show the content of A and B in their respective cluster. example is shown below:
      
______________     ->(100% A, 100% B)
|                          |
|                          |
|                          |
1                         2
(5% B)                (95% B)
(100% A)              (0% A)

In the first cluster (left branch) it shows all of A is in that cluster, similarly for 2nd branch majority of B is present in it. Thus we say that the signature are separable. If A was divided in equal ratios, then its separability won't be good.
I want to know, what you think about it and how this can be improved?

Thanks and kind regards,
Vishal



On 27 March 2013 01:10, vishal tiwari <[hidden email]> wrote:
i wasn't saying making a AOI seed pixel for an entire summer :P. It might be integrated with the current methods of selecting an AOI. Today when i was using the spectral processing extension, it can to my mind. 

thanks for the references.

Thanks and kind regards,  


On 27 March 2013 00:59, Adkins, Dustan <[hidden email]> wrote:

An AOI seed pixel is not nearly interesting enough for an entire GSoC project as it would take less than a week to implement.

I probably would not bother to implement it as the user can simply use the rectangle draw tool to do what you are suggesting.

 

 

As for the clustering, you will want to check out two of last year’s GSoC students, particularly the first one:

http://opticks.org/confluence/display/~himsin/Home

http://opticks.org/confluence/display/~mohitkharb/Home

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Tuesday, March 26, 2013 7:40 AM
To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

Thanks for replying.

I downloaded the spectral plugin, and saw how the signatures can be created, from an AOI or just using individual pixels. I was thinking a seed pixel can be used to select a number of pixels in its neighborhood by using one of the pixel neighborhood function(plane 4 neighbor or diagonal 8 neighbor), to create a signature. What do you think about it?

 

The dendogram, is just a tree showing the clustering results. In the spectral evaluation, we want to show how the clusters and divided/grouped until we have a desired number of cluster groups(more clusters or less clusters depending if clusters are grouped or divided). I think hierarchical clustering of signatures is required here. And a bottom-up approach seems more reasonable here. I want to know more about the clustering algorithm that was implemented last year.

 

Thanks and kind regards,

 

 

On 25 March 2013 16:38, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

 
This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d

_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs




--
Vishal Tiwari
Undergraduate
Computer Science and Engineering
Lab for Spatial Informatics
IIIT-Hyderabad



--
Vishal Tiwari
Undergraduate
Computer Science and Engineering
Lab for Spatial Informatics
IIIT-Hyderabad

------------------------------------------------------------------------------
Own the Future-Intel(R) Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest. Compete
for recognition, cash, and the chance to get your game on Steam.
$5K grand prize plus 10 genre and skill prizes. Submit your demo
by 6/6/13. http://altfarm.mediaplex.com/ad/ck/12124-176961-30367-2
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

dadkins
Administrator

Vishal,

 

That sounds like a decent start. I would like to take this idea and make it more general, e.g.:

 

Input:

Multiple spectral signatures or a pseudocolor layer (e.g.: as output by a clustering algorithm)

 

Output:

A graph displaying a hierarchical cluster (i.e.: dendrogram) which is grouped by similarity.

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Friday, March 29, 2013 1:12 AM
To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

i went through the ISODATA clustering algorithm, and the tutorial of its plugin at: http://opticks.org/confluence/display/~himsin/ISODATA+Plugin+Tutorial .

 

Here is a my idea for making the dendrograms: 

The initial number of clusters will be set to one. And as the algorithm proceeds, the number of clusters will increase. Now in each iteration, whenever the cluster splits, the dendrogram also splits, representing the two different set of clusters as the branches of dendrogram. The min standard deviation for split will be set to the max(standard deviation of the signatures) so that , once we have the signatures, it doesn't split further. It will be a top-bottom dendrogram. And at each level of the dendrogram at each branching we can have a node which will show the percentage/ratio of the count of a signatures pixel at that level. for example, if i have two signatures A and B, then at the top level, when there is no split(just one cluster) then it would display 100% A and 100% B, after an iteration, when the cluster splits in two clusters, each branch node will show the content of A and B in their respective cluster. example is shown below:

      

______________     ->(100% A, 100% B)

|                          |

|                          |

|                          |

1                         2

(5% B)                (95% B)

(100% A)              (0% A)

 

In the first cluster (left branch) it shows all of A is in that cluster, similarly for 2nd branch majority of B is present in it. Thus we say that the signature are separable. If A was divided in equal ratios, then its separability won't be good.

I want to know, what you think about it and how this can be improved?

 

Thanks and kind regards,

Vishal

 

 

On 27 March 2013 01:10, vishal tiwari <[hidden email]> wrote:

i wasn't saying making a AOI seed pixel for an entire summer :P. It might be integrated with the current methods of selecting an AOI. Today when i was using the spectral processing extension, it can to my mind. 

 

thanks for the references.

 

Thanks and kind regards,  

 

On 27 March 2013 00:59, Adkins, Dustan <[hidden email]> wrote:

An AOI seed pixel is not nearly interesting enough for an entire GSoC project as it would take less than a week to implement.

I probably would not bother to implement it as the user can simply use the rectangle draw tool to do what you are suggesting.

 

 

As for the clustering, you will want to check out two of last year’s GSoC students, particularly the first one:

http://opticks.org/confluence/display/~himsin/Home

http://opticks.org/confluence/display/~mohitkharb/Home

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Tuesday, March 26, 2013 7:40 AM
To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

Thanks for replying.

I downloaded the spectral plugin, and saw how the signatures can be created, from an AOI or just using individual pixels. I was thinking a seed pixel can be used to select a number of pixels in its neighborhood by using one of the pixel neighborhood function(plane 4 neighbor or diagonal 8 neighbor), to create a signature. What do you think about it?

 

The dendogram, is just a tree showing the clustering results. In the spectral evaluation, we want to show how the clusters and divided/grouped until we have a desired number of cluster groups(more clusters or less clusters depending if clusters are grouped or divided). I think hierarchical clustering of signatures is required here. And a bottom-up approach seems more reasonable here. I want to know more about the clustering algorithm that was implemented last year.

 

Thanks and kind regards,

 

 

On 25 March 2013 16:38, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

 
This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.

 

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d


_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad


This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Own the Future-Intel(R) Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest. Compete
for recognition, cash, and the chance to get your game on Steam.
$5K grand prize plus 10 genre and skill prizes. Submit your demo
by 6/6/13. http://altfarm.mediaplex.com/ad/ck/12124-176961-30367-2
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs
Reply | Threaded
Open this post in threaded view
|

Re: Opticks project in Gsoc 2013

vishal tiwari
For the more general way we would just do a normal hierarchical clustering of the signatures or the clusters that are the result of the some classification algorithm. In which a metric parameter will be taken like Euclidean Distance, Manhattan, SAM , or etc. This parameter will be calculated for each pair or signatures/classified clusters. And then the linkage criteria is selected( like max,min, or average), which will be used to group two signatures/cluster together. After each step, we will have at most n-1 clusters remaining(as similar valued pair can group together in a single level). Where n is the number of cluster in the previous level. On the dendrogram scaling factor will be shown, representing the metric parameter between the clusters. 

The method that i suggested before, can be used to see how an algorithm like ISODATA clustering divides the dataset into further clusters. But its drawback is, it might not give me the same number signatures at the end, two or more signatures might belong to the same cluster.
 
Maybe the user can select one of the methods to create the dendrogram.

Any thoughts on how it can further be improved?

Thanks,
Vishal


On 29 March 2013 16:51, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

That sounds like a decent start. I would like to take this idea and make it more general, e.g.:

 

Input:

Multiple spectral signatures or a pseudocolor layer (e.g.: as output by a clustering algorithm)

 

Output:

A graph displaying a hierarchical cluster (i.e.: dendrogram) which is grouped by similarity.

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Friday, March 29, 2013 1:12 AM


To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

i went through the ISODATA clustering algorithm, and the tutorial of its plugin at: http://opticks.org/confluence/display/~himsin/ISODATA+Plugin+Tutorial .

 

Here is a my idea for making the dendrograms: 

The initial number of clusters will be set to one. And as the algorithm proceeds, the number of clusters will increase. Now in each iteration, whenever the cluster splits, the dendrogram also splits, representing the two different set of clusters as the branches of dendrogram. The min standard deviation for split will be set to the max(standard deviation of the signatures) so that , once we have the signatures, it doesn't split further. It will be a top-bottom dendrogram. And at each level of the dendrogram at each branching we can have a node which will show the percentage/ratio of the count of a signatures pixel at that level. for example, if i have two signatures A and B, then at the top level, when there is no split(just one cluster) then it would display 100% A and 100% B, after an iteration, when the cluster splits in two clusters, each branch node will show the content of A and B in their respective cluster. example is shown below:

      

______________     ->(100% A, 100% B)

|                          |

|                          |

|                          |

1                         2

(5% B)                (95% B)

(100% A)              (0% A)

 

In the first cluster (left branch) it shows all of A is in that cluster, similarly for 2nd branch majority of B is present in it. Thus we say that the signature are separable. If A was divided in equal ratios, then its separability won't be good.

I want to know, what you think about it and how this can be improved?

 

Thanks and kind regards,

Vishal

 

 

On 27 March 2013 01:10, vishal tiwari <[hidden email]> wrote:

i wasn't saying making a AOI seed pixel for an entire summer :P. It might be integrated with the current methods of selecting an AOI. Today when i was using the spectral processing extension, it can to my mind. 

 

thanks for the references.

 

Thanks and kind regards,  

 

On 27 March 2013 00:59, Adkins, Dustan <[hidden email]> wrote:

An AOI seed pixel is not nearly interesting enough for an entire GSoC project as it would take less than a week to implement.

I probably would not bother to implement it as the user can simply use the rectangle draw tool to do what you are suggesting.

 

 

As for the clustering, you will want to check out two of last year’s GSoC students, particularly the first one:

http://opticks.org/confluence/display/~himsin/Home

http://opticks.org/confluence/display/~mohitkharb/Home

 

--Dustan

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Tuesday, March 26, 2013 7:40 AM
To: [hidden email]
Subject: Re: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

Thanks for replying.

I downloaded the spectral plugin, and saw how the signatures can be created, from an AOI or just using individual pixels. I was thinking a seed pixel can be used to select a number of pixels in its neighborhood by using one of the pixel neighborhood function(plane 4 neighbor or diagonal 8 neighbor), to create a signature. What do you think about it?

 

The dendogram, is just a tree showing the clustering results. In the spectral evaluation, we want to show how the clusters and divided/grouped until we have a desired number of cluster groups(more clusters or less clusters depending if clusters are grouped or divided). I think hierarchical clustering of signatures is required here. And a bottom-up approach seems more reasonable here. I want to know more about the clustering algorithm that was implemented last year.

 

Thanks and kind regards,

 

 

On 25 March 2013 16:38, Adkins, Dustan <[hidden email]> wrote:

Vishal,

 

It sounds like you have a pretty good grasp of what is going on here.

 

With regards to your questions:

 

I didn't understand how dendrogram can be used. Can you give me some guidance on this.”

Generically speaking, a dendrogram is just a tree structure used to visualize the results of a clustering algorithm.

Specifically for this project, you could use a dendrogram to organize signatures into something intuitive for the user (e.g.: break down types of vegetation).

This could include taking the results of a clustering algorithm (such as one of the ones written for last year’s GSoC) and displaying them in a tree structure for easy analysis.

 

“i wanted to ask, does the project also contain the part, for creating signatures?  “

Take a look at the Spectral Processing extension [1].

You would probably want to use the SDK provided with that extension in addition to the Opticks SDK for this effort.

 

Thanks,

Dustan

 

[1] http://opticks.org/confluence/display/opticksExt/Spectral+Processing

 

 

 

From: vishal tiwari [mailto:[hidden email]]
Sent: Saturday, March 23, 2013 4:52 AM
To: [hidden email]
Subject: [Opticks-devs] Opticks project in Gsoc 2013

 

Hi!

I have some questions regarding the projects of this years gsoc. I want to know if i am going in the right direction. I want your guidance regarding the projects.

 

This is what i have understood about the spectral signature tool:

1. To make a tool that will help the user/analyst to see/study the different signatures that have been created from an AOI.

2. The tool will consist of two ways of evaluation, one would be graphical and the other will be Tabular.

3. These two types can further have many subdivisions such as:

a) Graphical: 

It will contain the histograms of the signatures. And such histograms will be plot for different bands. Each signature can be differentiated from it.

Other graphical based evaluation can be Mean plots of the signature over several bands, or Scatter plots taking two bands at a time.

(I didn't understand how dendrogram can be used. Can you give me some guidance on this.)

b)Tabular: This will contain a matrix showing the value of one of the following:

Transformed Divergence, Jeffries-Matusita or Euclidean distances, between all the permutation of signatures that we have.

 

i wanted to ask, does the project also contain the part, for creating signatures?  

 

Please tell me if i am thinking in the right direction.

 

Thanks and kind regards,

 

-- 

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

 
This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.


------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

This message and any enclosures are intended only for the addressee.  Please 
 
notify the sender by email if you are not the intended recipient.  If you are 
 
not the intended recipient, you may not use, copy, disclose, or distribute this 
 
message or its contents or enclosures to any other person and any such actions 
 
may be unlawful.  Ball reserves the right to monitor and review all messages 
 
and enclosures sent to or from this email address.

 

------------------------------------------------------------------------------
Own the Future-Intel&reg; Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest.
Compete for recognition, cash, and the chance to get your game
on Steam. $5K grand prize plus 10 genre and skill prizes.
Submit your demo by 6/6/13. http://p.sf.net/sfu/intel_levelupd2d


_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad



 

--

Vishal Tiwari

Undergraduate

Computer Science and Engineering

Lab for Spatial Informatics

IIIT-Hyderabad

This message and any enclosures are intended only for the addressee.  Please 

notify the sender by email if you are not the intended recipient.  If you are 

not the intended recipient, you may not use, copy, disclose, or distribute this 

message or its contents or enclosures to any other person and any such actions 

may be unlawful.  Ball reserves the right to monitor and review all messages 

and enclosures sent to or from this email address.

------------------------------------------------------------------------------
Own the Future-Intel(R) Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest. Compete
for recognition, cash, and the chance to get your game on Steam.
$5K grand prize plus 10 genre and skill prizes. Submit your demo
by 6/6/13. http://altfarm.mediaplex.com/ad/ck/12124-176961-30367-2
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs




--
Vishal Tiwari
Undergraduate
Computer Science and Engineering
Lab for Spatial Informatics
IIIT-Hyderabad

------------------------------------------------------------------------------
Own the Future-Intel(R) Level Up Game Demo Contest 2013
Rise to greatness in Intel's independent game demo contest. Compete
for recognition, cash, and the chance to get your game on Steam.
$5K grand prize plus 10 genre and skill prizes. Submit your demo
by 6/6/13. http://altfarm.mediaplex.com/ad/ck/12124-176961-30367-2
_______________________________________________
Opticks-devs mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/opticks-devs