Does anybody have experience with using OpenCV in IgorPro 7?
mouthbag
Tue, 01/31/2017 - 02:41 am
I'm currently working on automating an experimental setup which incorporates a microscope which has to search for cell clusters and then record their positions. The preliminary goal is to have a program which can detect cell clusters on its own and record their positions over many hours. However, I am interested in implementing advanced image recognition in order to detect different types of cells and track their interactions.
Is there anybody who has experience with this kind of work? Can it be done in Igor?
Scince I have a C++ background, I thought about using OpenCV (or an equivalent image recognition package) and interface it with Igor. Does anybody have experience with this?
Best regards,
mouthbag
If you want to roll your own in C++, you need the XOP Toolkit for writing an Igor Pro extension [1]. If you know C/C++ writing an XOP is easy.
I would write operations as these are more flexibel than functions.
[1]: https://www.wavemetrics.com/products/xoptoolkit/manual.htm
January 31, 2017 at 08:18 am - Permalink
Current OpenCV is probably a magnitude greater in size. There is no question that it supports some algorithms which are currently not supported in Igor. If you are comfortable with C++ there is no reason why you could not create a nice XOP that supports the subset of OpenCV that you are interested in. As suggested above, the XOP Toolkit is your starting point. For Igor Pro 7 you should should probably focus on one platform and 64-bit configuration. I also agree with the previous post that operations are going to be more flexible here than functions. If you start with a single simple case of the form:
[OperationName] [Flags] [input waves]
which result in some fixed name output wave you would be able to quickly duplicate the code to add support for a variety of OpenCV functions.
A.G.
January 31, 2017 at 11:19 am - Permalink