Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

Java Programme

Options
  • 15-06-2005 7:04pm
    #1
    Closed Accounts Posts: 295 ✭✭


    Has anyone got a Java programme that represents the image taken from a Camera. It will need to take into account the distance between a point and camera and intensitity


Comments

  • Registered Users Posts: 2,426 ✭✭✭ressem


    Guess that this is related to your other qn.
    Looking for a library/program that uses edge detection and the like, to map a 2D image into a 3D model is probably necessary.
    Unless you're using photos of a lake/infinite flat plain/ uniform brightness / some sort of reference (ruler/ flat road/ equidistant street lights) in the picture that can be read by your program ?

    Something like
    http://www.realviz.com/products/im/howitworks.php
    http://www.photomodeler.com/
    http://www.3dcgi.com/cooltech/modeling/image-modeling.htm

    If you want a free version, I hope your maths is up to scratch.


  • Closed Accounts Posts: 295 ✭✭bullock


    Yes Those sites were a big help. But its the Java code thats behind www.realviz.com that I would need. The idea of a camera taking too shots of the one area is the general idea.


  • Registered Users Posts: 2,426 ✭✭✭ressem


    :rolleyes:
    I'll bet you do.

    1. Why does it need to be a java library? Is JNI out of the question? I glanced at that monitor and the provided app listed uses DirectX to display 3D models.
    2. Are you able to purchase the libraries/ use proprietary code/ use gpl code?
    3. Does it have to be fully automatic?
    4. How fast does it have to work?
    5. How much time/money are you willing to put in?


    Depends on what your images are of. (IMO it would help if you would elaborate more than a line at a time)
    eg. for medical stuff, where your images are from an MRI and you want to display using Java3D, you could have a look at http://www.quickvol.com/
    License is for research only so no codestealing.

    Or have you a setup like http://amp.ece.cmu.edu/projects/MobileCamArray/

    Other Libs
    http://www.tina-vision.net/
    Provides the 3D matching tool from 2D images apparently. Requires permisson to use commercially.

    http://www.mvtec.com/products/
    Looks like some of the functions in Halcon Tools/ could do what you want.


    Reading:
    http://homepages.inf.ed.ac.uk/cgi/rbf/CVONLINE/entries.pl?TAG1076

    http://www.intel.com/research/mrl/research/opencv/
    http://sourceforge.net/projects/opencvlibrary
    (howto + course)

    Course points out using Hough transforms to do line detection, which might be useful to give your program an idea of the change of perspective.

    Then an approximate partial 3d model (using textures extracted from the image) has to be built.


  • Closed Accounts Posts: 295 ✭✭bullock


    I just need a bit of Java code. This is what it is. I'm looking to start a piece of Java code that will allow me to take into account the rotation and translation Matrix of the Camera. I must also take into account the distance from the point of interest to the centre of the camera ( Camera Offsets) as well as the camera focal length.

    Lets say that camera a takes an left image and camera b takes an image of the right image. I am very unclear of how I am going to start it and what values to take for my Camera focal length and Offsets.

    Hopefully I have made it some way clearer and someone can help get me started.


  • Registered Users Posts: 2,426 ✭✭✭ressem


    Sorry, havn't seen any Java libs other than academic bits that go with a thesis.

    If I was doing this from scratch,
    I'd experiment with.
    1 running a canny filter to do edge detection.
    2 running the hough transform to do line segment detection on the result.
    3 try to match the line segments on each picture, attempting to overlay the images
    4 Use basic tech graphics/optics maths to try and resolve where the cameras lie based on these lines.

    Or put restrictions on the camera setup, eg stereoscopic setup required, (Or ask the boss to stump up for a startech 3d rangefinding camera for 20G :->)

    But the C, C++ libs listed would make this a lot easier. So I'd be trying to call the C libs through COM, JNI, corba.

    Especially if you need to actually draw this thng in 3d.

    I'll shut up now and let someone else take a shot.


  • Advertisement
Advertisement