3

I am working on a project to monitor if human tissue has been fused with radio frequency during the surgery or not, therefore we are using a very fast camera (1800fps) and also laser illumination on the tissue and a framegrabber (1GB memory).

At the moment we have a software based on C which is only getting the data from the grabber and passes it on the PC also show the images in Screen but I need to add an algorithm to grab all the frames from the framegrabber and analyze them with a mask and also calculate the contrast deviation of temporal images in real time.

I think there must be a framework out there which help me for the image analyzing in C++. What do you suggest? Zhat would be the best option for me to do the processing as fast as possible in real time ? OpenCV? OpenGL? GLPlus?

CharlesB
  • 86,532
  • 28
  • 194
  • 218
user261002
  • 2,182
  • 10
  • 46
  • 73

2 Answers2

3

to answer your question I think you should look into OpenCV and ITK, the second one being somehow more popular in the medical imaging community.

That being said, what kind of processing do you intend to do? At these framerates you can only do simple things. I know that some frame grabbers can preprocess images for you and that might be enough for your application. IMO GPU computing is out of the question because of the data transfers and the simplicity of your computation

Adrien
  • 146
  • 7
  • 1
    VTK is a visualization toolkit, it is not made for images analysis. – CharlesB Mar 21 '12 at 12:39
  • I'm kind of new here, is there a prefered way to edit a post? Should I make it clear that i made a mistake (like i did), or should I just replace VTK by ITK in the post? – Adrien Mar 21 '12 at 16:28
  • just replace VTK with ITK, edits are logged in a history so you don't have to *tell* you edited something, unless it adds new stuff to your post. – CharlesB Mar 21 '12 at 17:02
2

OpenCV Is your buddy for the tasks that you want to achieve. It basically supports lots of image processing functions, supports reading videos via FFMPEG and has camera capture options too. With OpenCV on TBB and GPU enabled primitives, you can definitely harness GPU power for the sake of speed. Plus tonnes of community contribution is easily available as OpenCV is slowly but surely becoming a MUST know tool for image analysis experts/developers.

You might go for OpenGL. But it would suit 3D graphics type of projects more than just processing videos/image/frames.