#1
  1. No Profile Picture
    Junior Member
    Devshed Newbie (0 - 499 posts)

    Join Date
    Aug 2003
    Posts
    7
    Rep Power
    0

    Question image processing


    1)
    how can i dettect movement through image processing and get following information
    * size of the moving object
    * direction of motion

    2)
    how can i get the "car number"(registration number,number plate) through image processing

    thanks
  2. #2
  3. Doggie
    Devshed Novice (500 - 999 posts)

    Join Date
    Jul 2003
    Location
    Seattle, WA
    Posts
    751
    Rep Power
    13
    Could you be a little more descriptive? What program/language are you working with? What do you mean by "movement through image processing"?
  4. #3
  5. No Profile Picture
    Junior Member
    Devshed Newbie (0 - 499 posts)

    Join Date
    Aug 2003
    Posts
    7
    Rep Power
    0

    Question re :image processing


    hi
    here is brief introduction of my project (border security system)

    HOW IT WORKS:-
    The system provide the three dimensional security through the “three” cameras. Cameras are to used to capture movement inside the restricted border area.The cameras are working according to the scheduled criteria.The camera pass the image(snapshot) to the Pc in the sub-control office. On the Pc side different algorithm are applied to the most recent two images to detect and extract motion and other information e-g direction,speed,size of the object.

    If no motion detected the process is repeated for each camera.

    The system generate the “angry barking dog” sound after random interval of time in the night.

    After motion detection in the restricted area the system perform the following thing:-

    i) PC SIDE:-
    After snapshots sended to the PC and on motion detection :-
    1) Show the following information on GUI:
    * Area of motion
    * Distance from the sub-control office
    * Speed of the object
    * Size of the object
    2) Send signal to other sub-control office.
    3) Save information to the database for future forecasting.
    4) Save images to the disk for future analysis .

    ii) MICROCONTROLLER SIDE:-
    The response of microchip is depended on two things
    * Direction of object
    * Size of the object
    The information pass to microchip and it responses the following thing according to that information
    1) Activate the alarms in that control area.
    2) Activate the search light (Night) .
    3) Make a telephone call to the main control office and deliver the
    recorded message.



    may want to use visual basic.
    Last edited by daniahmed; August 5th, 2003 at 01:37 AM.
  6. #4
  7. Doggie
    Devshed Novice (500 - 999 posts)

    Join Date
    Jul 2003
    Location
    Seattle, WA
    Posts
    751
    Rep Power
    13
    Well, I wouldn't use Visual Basic for this. I've made graphic programs with VB and they're very slow, no mater how much I optimized them.

    This is pretty easy to do, as long as the cameras aren't moving. (ie: panning accross an area)

    Comparing your new image to your reference image (where you know there's no movement) find all the pixels where the color (or average in an area) has changed more then a defined amount. Make that your mask.

    You may come up with multiple objects in your mask. (areas of concurent pixels) You'll want to remove objects that have fewer then a defined number of pixels. Those will probably just be noise between the images.

    Calculate the bounding box of the objects. (top, bottom, left, right) That will give you your general size. You could also just sum up the number of pixels in the objects.

    Between two masks, calculate the difference between the centers of two bounding boxes to calculate speed. If more then one object in the mask, match objects between images by similar sizes and proximity to each other.
    Last edited by dog135; August 5th, 2003 at 12:35 PM.
  8. #5
  9. No Profile Picture
    Contributing User
    Devshed Newbie (0 - 499 posts)

    Join Date
    Feb 2002
    Location
    BCN
    Posts
    84
    Rep Power
    13
    Yes, do not use VB. First of all because if you want to do some low level manipulation I would suggest C or C++.
    For what concerns the algorithmics, I think you should search info about image processing using tools as Fourier analysis. I would also suggest that, if the surveyed area is always the same you provided the system with some "quiet" image (I mean, what you expect to see when looking to a certain direction) to compare against . This way you could know where something strange is happening and how is it changing.
  10. #6
  11. Doggie
    Devshed Novice (500 - 999 posts)

    Join Date
    Jul 2003
    Location
    Seattle, WA
    Posts
    751
    Rep Power
    13
    Originally posted by ygdrasyl
    I would also suggest that, if the surveyed area is always the same you provided the system with some "quiet" image (I mean, what you expect to see when looking to a certain direction) to compare against.
    The problem with that, is natural lighting. If there's a skylight or window anywhere nearby, then the change in lighting over time would make the whole image change relative to the reference image.

    To get around that, when you get two consecutive images without any detectable objects in them, use the newer one as your new reference image.

    Other notes:

    What format are these images in? JPGs and GIFs take too much time to decompress and often have artifacts so you'll want a more raw format.

    How often are you receiving these images? If your delay between images is a second or more, you're likely to be unable to determing speed and direction because the movement will be to fast. If over a second between images, you could miss the object all together.

IMN logo majestic logo threadwatch logo seochat tools logo