Do PixelValue2D properties return a new array, or a reference?

[middletongard]

I'm writing a program to acquire data from an IMAQ device to a ring buffer, and operate on the data from there. I'm extracting a PixelValue2D, and I'm curious about how it works.

I want to access the data as a short[,], so I write:

PixelValue2D pixVal = _session.Acquisition.Extract(bufferNumber, out bufferNumber).ToPixelArray();

short[,] imgData = pixVal.I16;

Does the pixVal.I16 return a reference to the data, or a new copy of the data? This information will be very helpful for keeping my program as efficient as possible.

Thanks!

[Sarci1]

Hi middletongard,

Could you tell me which language you are working in so that I may be of better assistance?  I have not been able to find a reference to this function.

Lynn

[middletongard]

C#

Thanks

[middletongard]

I found information for PixelValue2D.I16 here within the Visual Studio 2008 help:

ms-help://MS.VSCC.v90/MS.VSIPCC.v90/NI.MeasurementStudio.2008/NINETVision/html/NationalInstruments.Vision.PixelValue2D.I16.html

PixelValue2D.I16 Property

PixelValue2D Class | PixelValue2D Members | NationalInstruments.Vision Namespace

Language

• C#
• Visual Bas All

Gets the pixel values for images of type I16.

[Visual Basic]
Public Property ReadOnly I16 As Short(,)

[C#]
public short[,] I16 {get;}

Property Value

The pixel values for images of type I16.

See Also

PixelValue2D Class | PixelValue2D Members | NationalInstruments.Vision Namespace

[Sarci1]

Hey!

Ah yep, I found that right after you.  Well I'm glad you've got a solution. Best of luck with your application!

Regards,

Lynn

[middletongard]

Ah, I don't have the solution quite yet - the information does not tell me whether a copy of the data is made, or a reference to the data is returned.

If this information is implied in the definition, I haven't caught it. Could you please find out and let me know explicitly whether a new copy of the data is made and returned, or it's simply a reference to the data in PixelData2D (no copy) that is returned?

Thanks

[Sarci1]

Hi middletongard,

It appears as though the PixelValue2D is ImaqBuffer which a a zero-copy operation.

Lynn

[OrangeAndGreen]

Similar to middletongard, I'm trying to acquire IMAQ data in C# and extract the data into a short array for further processing. From the last post I see that extracting PixelValue2D from ImaqBuffer is a zero-copy operation, but I still wonder whether new memory is being allocated?

The reason I ask is that my application crashes with an OutOfMemoryException after running for a couple seconds.

This is the line of code that generates the exception:

short[,] data = mSession.Acquisition.Extract(BufferNumber, out CurrentBufferNumber).ToPixelArray().I16;

And this is the relevant part of the stack trace:

Exception Info: System.OutOfMemoryException

Stack:

   at NationalInstruments.Vision.Internal.Utilities.ConvertIntPtrFlatTo2DArrayInt16(IntPtr, Int32, Int32, Boolean)

   at NationalInstruments.Vision.VisionImage.ImageToArray(NationalInstruments.Vision.RectangleContour)

   at NationalInstruments.Vision.Acquisition.Imaq.Internal.ImaqVisionImageBuffer.ToPixelArray()

   at NationalInstruments.Vision.Acquisition.Imaq.ImaqBuffer.ToPixelArray()

Note that the exception is thrown from ToPixelArray(), not getting the I16 property. 

I would appreciate any help I can get!

[BlueCheese]

Accessing raw pixel data as an array out of a Vision image is always a copy operation. Otherwise there is no way to properly synchronize access between the Vision image's access to its internal data and the user's access via the array.

Eric

[OrangeAndGreen]

Thanks for the quick response Eric, that definitely makes sense to me. But my issue is with the memory allocation rather than the copying. I would like to reuse memory instead of repeatedly allocating and freeing.

A desireable solution would be if I could pass in the destination array when extracting the data, i.e.:

//During initialization

   short[,] data = new short[xSize, ySize];

//Later, during acquisition

   mSession.Acquisition.Extract(...).ToPixelArray(out data);

I know that call doesn't exist, but I wonder if there is a workaround for it.

My application can't afford to run the garbage collector during acquisition, so the goal is to allocate buffers before the start of acquisition and then reuse them.