[Insight-users] Re: DeformableRegistration8.cxx std::bad_alloc

[Luis Ibanez]

Hi Ruben,

Thanks for the detailed description of the problem you are facing.

There are many things that go on in the StartRegistration() method.
Actually, all of the registration process happens during that call.

We will need to track the problem by going inside the method and
locating the exact place where the exception is thrown.

Is it possible for you to run the program from a debugger ?

If you look at the implementation of the StartRegistration() method
in line 230 of:


       Insight/Code/Algorithms/itkImageRegistrationMethod.txx

You will see that the main actions in this methods are

   1) a call to Initialize()
   2) a call to StartOptimization()

If you can identify which one of these two method throw the
exception, we could help on further locating the source of the
problem.


   Please let us know what you find.



       Thanks


          Luis



---------------------
Ruben Schilling wrote:
> Hi again,
> 
> In the meanwhile I have located the LOC, where the program aborts, it  
> is the "registration->StartRegistration()" call in this try block:
> 
> 
> std::cout << std::endl << "Starting Registration" << std::endl;
> 
> try
>     {
>     collector.Start( "Registration" );
>     registration->StartRegistration();
>     collector.Stop( "Registration" );
>     }
> 
> 
> 
> Does anyone have suggestions why this could be a reason to throw a  
> std::bad_alloc exception? Could there be a problem with some  
> dynamically loaded ITK function? What would I need to do then? My  
> CMakeLists.txt looks like this (setting the variables was necessary  to 
> compile and link everything):
> 
> 
> 
> PROJECT(DeformRegister8)
> #set paths to find ITK
> SET(ITK_DIR /project/algorithmics/ruben/bin-x86_64/lib/InsightToolkit)
> SET(ITK_SOURCE_DIR /project/algorithmics/ruben/bin/ 
> InsightToolkit-3.0.0/x86_64)
> SET (CMAKE_INSTALL_PREFIX /project/algorithmics/ruben/bin-x86_64)
> 
> FIND_PACKAGE(ITK)
> IF(ITK_FOUND)
>   INCLUDE(${ITK_USE_FILE})
> ELSE(ITK_FOUND)
>   MESSAGE(FATAL_ERROR "Cannot build without ITK.  Please set ITK_DIR.")
> ENDIF(ITK_FOUND)
> 
> LINK_LIBRARIES (
> ITKNumerics ITKFEM ITKIO ITKStatistics
> )
> 
> ADD_EXECUTABLE(DeformableRegistration8 DeformableRegistration8.cxx )
> INSTALL_TARGETS(/bin DeformableRegistration8)
> 
> 
> 
> 
> 
> Am 17.03.2007 um 19:52 schrieb Ruben Schilling:
> 
>> Hi,
>>
>> I am using the DeformableRegistration8.cxx example to test  
>> registration on image volumes using Mutual Information. I have a  
>> problem though:
>>
>> The program crashes with this error message
>> >>>>>>>>>>
>> Starting Registration
>> terminate called after throwing an instance of 'std::bad_alloc'
>>   what():  St9bad_alloc
>> Aborted
>> <<<<<<<<<<
>>
>> From the message I can see, that the application ran acutually at  
>> least until line 359 in the Code of DeformableRegistration8.cxx
>>
>> The volumes, that I try to register are large: 8bit grayscale, but  
>> about 400x500x300 pixels in Volume each. They are stored as Tif  
>> stacks. I am aware, that this requires a lot of memory, but I have  
>> compiled the DeformableRegistration8.cxx and all of ITK etc. as  64bit 
>> (see output of 'file' command)
>>
>> ELF 64-bit LSB executable, AMD x86-64, version 1 (SYSV), for GNU/ 
>> Linux 2.6.0, dynamically linked (uses shared libs), not stripped
>>
>> The Linux computer I use has 64 Gigabyte of RAM (and I am currently  
>> the only user on it), thus I can't believe I would not have enough  
>> memory at all. So what's the matter with DeformableRegistration8.cxx?
>>
>> Can anybody help? Any suggestions? Does the implementation of  
>> MutualInformation have numerical problems? Actually from the error  
>> message I would have assumed, that it is an error related to the  
>> 'new' operator. Looking with top at the memory usage using a fast  
>> update, I had seen, that the registration just crashed after  
>> assigning about 300-400 MByte of memory.
>>
>> I would be grateful (and curious about the solution) for any help  on 
>> this matter, since I really need deformable multi-modal  registration 
>> on this data.
>>
>> Ruben
> 
> 
> _______________________________________________
> Insight-users mailing list
> Insight-users at itk.org
> http://www.itk.org/mailman/listinfo/insight-users
>