tex and i have been discussing strategies to get OpenCV to processing / java. right now there are several approaches.

+ JNI: manually write a JNI binding to the OpenCV interface.
+ SWIG: use SWIG to create a native binding. with the perspective of using the python SWIG files that create the 'official' python binding.
+ OSC: write a 'binding' to OSC either in C++ or in java ( need JNI again ) much like eyesweb.

right now we are in the process of investigating the feasibility of these approaches.

we would really like to start a project on this. since we don t have that much spare time at hand, the progress is very slow though. but opencv is really promising and definitly worth the effort.

great.
i just registered a project at sourceforge.

  OpenCV for Processing

this is the proposal i wrote to sourceforge. it roughly outlines the roadmap.
Quote:


i probably won't work on it until the end of the month. but i would be absolutely happy to see people joinging the project.

the first task for me would be to investigate the existing SWIG interface, currently used for the python binding and see what can be used for the low level java binding.

in terms of tools, we will probably use a combination of eclipse plugins: java, cdt, skwash. although their value still needs to be validated and is part of the developement process.

Good afternoon.

Are you done with the JNI binding of potrace?

Since this is excatly what I was looking for my question is, do you plan to share or publish it?  

If so I am really interested in it.

trace

This is all very exciting, as I was just looking into OpenCV as a possibility, but outside of Processing.

JNI isn’t out of the question for us, but we’ll have to look at the project more to see what we can contribute. More later...

does anyone made a version for PC or even Linux ?

Yeah, we're almost there. Mac OS (Universal Binary) works. Just finishing documentation and cleaning up things. We'll release some time in the next few weeks. We need PC compatibility and need Linux compatibility and there's no reason these shouldn't work. But no guarantees until we're done.

Unfortunately, don't expect the full 100% of OpenCV. So far, we only have edge detection and basic image manipulation with some filtering. We'll add more as we evolve. Also, we'd like some feedback from the Processing community on what people want to see from the OpenCV base, and also if anyone understands what the §@#& those mathematical formulas in the documentation mean. I'm just not a math nerd, sorry.

On the positive side, you can use OpenCV to capture directly from your camera, and you can also feed it your own image. So you don't need to use the Processing capture class in that case. Or you could feed it webcam images from the web, or your own sketch window, whatever. You can also get those images back after they've been worked on. I.e. OpenCV can be used as a basic image processor. So that's cool.

On the downside, OpenCV capture only works on a few webcams (iSight, for example) and only at 640x480. We can quick-shrink it for you immediately after capture so that your analysis goes faster, but it's still pulling it off the camera at 640x480. This is probably not much of a problem, but still. We find the edge detection at 640x480 to be perfectly workable and fast enough for most needs (with the added advantage of high resolution), but you have the option nevertheless.

Oh, and no parameter settings for your camera yet, which totally sucks. But both of those are OpenCV-port-to-Mac problems, so now we're going to have to work on the original source to fix those problems. Again, we still have a long way to go.

tibou wrote on Jan 19^th, 2008, 7:57pm:


We'll be working on PC this weekend, and trying to get it working over the next two weeks during a residency. Should have something at the end of all that.