Composite Video Cropping IC

Discussion in 'The Projects Forum' started by dnlwltn, Jul 31, 2013.

  1. dnlwltn

    Thread Starter New Member

    Jul 31, 2013
    5
    0
    Hi everyone. I'm in the process of developing a camera application for my company. I have searched high and low for an IC that provides the ability to crop analog Composite video to a variable size and position. I am also open to using software in a PIC microcontroller to accomplish the same, but I am sure there has to be an I2C controlled IC out there that already does this. Please help.
     
  2. Sensacell

    Well-Known Member

    Jun 19, 2012
    1,128
    266
    What your asking for is very difficult and complex.

    There is a chip, the Micronas SDA95

    Just looking at the data sheet makes my head ache.
    The chip can probably do the job, but implementing the software and hardware supporting the chip is still a massive task.

    You might ask Micronas for a evaluation board, or a reference design as a starting point.
     
  3. BReeves

    Member

    Nov 24, 2012
    412
    64
    Machine vision cameras can do what you are asking in the camera. It's refered to as ROI or "Region Of Intrest" and can be configured on the fly if need be. Look up AVT cameras for an example.. You will need a PC but a laptop can be used.
     
  4. cornishlad

    Member

    Jul 31, 2013
    196
    25
    I'm intrigued by the question and don't understand it :) Sensacell thinks you want to do picture in picture, BReeves thinks you want to do some sort of selective compression. Help us out with more details....
     
  5. dnlwltn

    Thread Starter New Member

    Jul 31, 2013
    5
    0
    Thanks for the responses guys. I will look into the micronas chip and see if a search using the term ROI help. To clarify what I am looking for: I have a composite video signal. I need to take one region of the video and expand that region to full screen. Think of it as digital zoom, but on a composite video signal instead of a digital signal.
     
  6. rogs

    Active Member

    Aug 28, 2009
    279
    37
    As Sensacell says, this is a complex task - and if you're trying to do this in real time, then that's quite a challenge!

    If however, you can record the original video, then cropping and resizing on a computer, and displaying the resultant video stream in real time, without re-encoding, is a pretty simple task.....although it will involve digitising the composite video when initially recording it onto a computer of course.....and the amount of 'cropping' and 'zooming' will affect the quality of the final video......it is in effect a 'digital' zoom....

    Are you expecting to do this real time, or can you record the video, and replay later?.......
     
  7. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    Edit: Oops... I keep on posting in the wrong thread!
     
    Last edited: Aug 2, 2013
  8. dnlwltn

    Thread Starter New Member

    Jul 31, 2013
    5
    0
    I am trying to zoom real time. I need the input AND output to be composite (cvbs) video. I want to do it at the chip level so that my application can make real time adjustments via serial commands. Using a computer, unfortunately, is not an option. All good answers and good questions to think about so far. Thanks for the help.
     
  9. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    I believe this is not a simple thing to design and build. Go out and buy a multi-camera surveillance system.
     
  10. rogs

    Active Member

    Aug 28, 2009
    279
    37
    I really can't see how you're going to do this......but if you do find a way, please let us know how......I for one would be fascinated!..
     
  11. cornishlad

    Member

    Jul 31, 2013
    196
    25
    you want the zoomed output to be full screen ? And the human interface is a joystick or something ?

    and what is the resolution of the input signal: broadcast 625 standard..low res security camera..etc ?

    colour or monchrome ?
     
    Last edited: Aug 1, 2013
  12. cornishlad

    Member

    Jul 31, 2013
    196
    25
    I'll answer myself ! no doubt you will be back...
    It's 20 years since I worked in broadcast tv and we did have a limited ability to do that with a comprehensive vision mixer (big bucks) We could only zoom in concentrically to eliminate a small amount of picture margin.
    I'm sure you realise that composite is an anologue signal and it will have to be processed digitally. ie...A>D...Digital processor >D>A
    I see a 2U rack ! I doubt very much that is a one chip solution. If the processor box doesn't exist ready made I also see weeks of R&D..
    I would add that, in the absense of a reply to the questions, if your comp vid is low res the effect of zooming to any usefull degree will be questionable. Only in CSI can they zoom to a number plate and "enhance" it..
     
  13. rogs

    Active Member

    Aug 28, 2009
    279
    37
    When I saw your questions in post #11, I really thought you were going to suggest a brilliant 'analogue' answer that the rest of us had missed! :)

    I think this probably a bit more complex than the OP realises.....
     
  14. cornishlad

    Member

    Jul 31, 2013
    196
    25
    I just wanted to make absolutely sure we had got the full picture and had not missed something ! As it hung around unanswered I must admit I can understand how you felt re the brilliant solution..
    But I also thought that anyone who had the skill/knowledge to execute such a processor wouldn't have posted the question !
    The OP did say in #1 that he was "developing it", implying it was under way so thought caution was called for in appearing to be too negative.
    I hope he will come back and report progress.......
     
  15. John P

    AAC Fanatic!

    Oct 14, 2008
    1,632
    224
    I hate to use the word "impossible", but this just about deserves it. What you have to consider is that the visible part of a TV screen has 483 lines. If you want to zoom vertically, you have to work in multiples (1:2, 1:3 etc--you have to receive a line once, then play it back multiple times) of a given line and you also have to have some facility for holding the contents of the line and playing it back 2 or 3 or 4 or some number of times. Note that while (let's say) line 69 is being shown several times for an enlarged view, line 70 and 71 and 72 will be delivered by the incoming video line, and you'll have to store those too, before you play any of them a single time.

    Realistically, this can only be done with digital hardware, and it is quite a complicated process. There are probably commercial devices that can do it, but I wouldn't care to try designing such a thing.
     
  16. rogs

    Active Member

    Aug 28, 2009
    279
    37
    Another confirmation of some of the posts above, about how difficult this would be to achieve without a major R&D budget......
    ....I think the OP is probably getting the message by now :) .....

    dnlwltn..... probably the only realistic way of achieving what you need to do, as a 'one off' project, is to utilise a camera (or camcorder) that has an internal 'zoom' facility, and use a serial control of that zoom function (via a LANC controller or similar) to control your video optically....or, if the camera has a digital zoom, to use that as a less optimal alternative.
     
  17. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    The simplest solution is to put a video camera in front of a display screen and control the pan and zoom of the camera.
     
  18. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,006
    3,763
    This is just not a hobbyist project or a single board project if real time analog in/out is required. I am not sure this request is even possible (enomicallyfeasible).

    If you really need this to happen and a possible solution is: composite-to-digital-to-composite, then contact someone like...

    http://www.compunetix.com/ix/solutions_technology.php
     
    Last edited: Aug 3, 2013
  19. MrChips

    Moderator

    Oct 2, 2009
    12,432
    3,360
    All joking aside, what you need to do is digitize and capture the picture into a frame buffer and then recreate the composite signal using the selected ROI (region of interest).
     
  20. cornishlad

    Member

    Jul 31, 2013
    196
    25
    Some nostalgia !....When I first worked in TV in 1961 - the days of 405 monochrome and valves, I was an assistant in a studio maintenance workshop. For a drama production that was planned an engineer on our shift came up with an idea that was dubbed "the dream effect". He fed some audio tone into a camera line scan board causing the picture to have "waves" of horizontal picture displacement running down the screen, triggered gently by a "volume control"
    The effect was wound in at the critical point, the camera defocused, cross mix to another defoccused camera which then focuses on the new scene (of the dream). Everyone was stunned at this inovation and much patting of backs ensued...
    Now, more than 50 years later what we see on out screens is limited only by imagination, expert skills and processing power. What the OP wants to do can be done by all of us for one frame, with a couple of clicks in Photoshop. Apart from the supporting hardware it is software process.
    An earlier post mentioned doing it "post production" in an editing program. Many of these programs can now do amazing effects but often they have to be "rendered" which as I'm sure you all now, means leaving the PC for some time to calculate and produce every frame. However, with a powerful PC supported by a graphics engine (card) much can be done on the fly.
    I'm sure that what our OP wants to do is not impossible in general terms but much more will be needed than his undoubted ability to programme a microprocessor. Apologies for the ramble :)
     
Loading...