VideoHelp Forum
+ Reply to Thread
Results 1 to 10 of 10
Thread
  1. Member
    Join Date
    Dec 2015
    Location
    world
    Search Comp PM
    Hoping I am in right forum. I have about 400 cameras in the cctv. Wondering if there is a software that would analyze the video for me (such as whether the camera is out of focus, camera is not working, frame rate is low etc).
    Quote Quote  
  2. Member Cornucopia's Avatar
    Join Date
    Oct 2001
    Location
    Deep in the Heart of Texas
    Search PM
    With the exception of "not working" those are all human-perspective qualitative judgements, so no software is going to truly give you what you're looking for.

    Scott
    Quote Quote  
  3. Member
    Join Date
    Dec 2015
    Location
    world
    Search Comp PM
    Out of focus is pretty well software defined, as some cameras have autofocus. Low frame rate is also monitored on some cams.

    Is there any software that could capture the video in java and analyze it to some extent?
    Quote Quote  
  4. Even if one could code such a piece of software do you have any idea how powerful a computer you would need to simultaneously check 400 video feeds and adjust the cameras automatically? And that's assuming that the cameras allow for adjustment via third party software.

    As for "capturing the video in Java" that statement doesn't make any sense, I assume you mean you want a program written in Java that is capable of capturing 400 video streams and then checking them to see if they fit certain predefined parameters.

    LOL! That's pretty funny, thanks I needed a laugh.
    Quote Quote  
  5. You poll them one at a time, obviously.
    Quote Quote  
  6. Member
    Join Date
    Aug 2006
    Location
    United States
    Search Comp PM
    Originally Posted by koolkool View Post
    Out of focus is pretty well software defined, as some cameras have autofocus.
    Either you have a no idea how autofocus cameras really work, or you are deliberately pulling our leg.
    Quote Quote  
  7. Focus is pretty easy to detect in software. You just run the picture through a high pass filter. What's left are sharp edges and noise. The noise is easily ignored via a threshold value. This isn't perfect (in the same way autofocus on camera's isn't perfect) but since security cameras are usually looking at fairly static scenes you know what to expect from each one.
    Quote Quote  
  8. Member Cornucopia's Avatar
    Join Date
    Oct 2001
    Location
    Deep in the Heart of Texas
    Search PM
    FOCUS:
    This is a descriptive thing, not a prescriptive thing. How does a software know something is in focus? - It doesn't. It only knows 2 things: metadata telling it where the focus is "supposed" to be, and pixel data that can let you know which areas are sharp and which aren't. It doesn't CARE if the focus is closer or farther than YOU'D like it to be, it can only tell you what areas it shows as sharp, and unless you have a multi-sensor camera those areas don't tell you anything about DEPTH, which is the most important detail regarding focus. Remember, it doesn't KNOW what your intended "subject" is supposed to be. It also doesn't know nor tell you WHY those pixels are sharp or not. Certain things can be guessed at based on additional data, but there are a number of types of "blur" that look similar even though they are created from different influences. Look at how hard it is to artificially generate good BOKEH.

    AUTOFOCUS:
    Cam metadata can tell you if it's on, or off. Can't tell you that you are using occasional manual adjustment. Can't tell you WHAT ACTUAL SUBJECTS OF INTEREST are currently being the primary targets of focus adjustment, it can only tell you which regions of the sensor screen are weighted most strongly (ROI/POI - Region/Point of Interest) and (if sonar/IR/etc. ranging is used) what the current focal depth is estimated as. No correlation between that and what your intended target of autofocus might be.

    WHITE BALANCE:
    Same problem as focus. Metadata from the cam might tell you that the WB is optimized for 3200ēK or 5600ēK, etc. Not that that is what it OUGHT to be set for, because that requires knowledge of the combination of existing lighting in that scene and the desired look. It doesn't express "the best of multiple bad options" either, for those cases where there is varying or multiple different light sources.

    EXPOSURE:
    Again, it can tell you if you are making use of certain ranges of pixel light values, but it cannot tell you whether the range recorded is being FAITHFUL to the scene (watch out, high-key and low-key!). And it cannot do anything about scenes that exhibit higher dynamic range than the cam is able to accommodate - something is bound to be clipped/crushed and you have no control over that other than choosing which scale end(s) it will affect (highlights, shadows, or both).

    FRAMERATE:
    Not counting very low quality cameras and phone cams, this will be constant once you set it, so there's really no need to "monitor" it. Asking "is it too LOW of a framerate?" is really only half of the question, because the other half is "...for the motion of the subject that I'm interested in and the look that I want to convey...". No software can describe that to you, much less prescribe.

    EXIF-type metadata from the cam is the best information source you will have available, but even that is ONLY superficially descriptive (not prescriptive at all) and could even be faulty/mistaken (as they are never really calibrated).

    Scott
    Last edited by Cornucopia; 23rd Dec 2015 at 12:32.
    Quote Quote  
  9. Member
    Join Date
    Aug 2006
    Location
    United States
    Search Comp PM
    Originally Posted by jagabo View Post
    Focus is pretty easy to detect in software. You just run the picture through a high pass filter. What's left are sharp edges and noise. The noise is easily ignored via a threshold value. This isn't perfect (in the same way autofocus on camera's isn't perfect) but since security cameras are usually looking at fairly static scenes you know what to expect from each one.
    Your idea would not work unless the camera is indoors and the lighting never changes.

    ...but autofocus doesn't work like that. Autofocus usually works by analyzing data from special sensors built into the camera, not picture information. Frequently the sensor data is used to measure distance to the nearest object seen though the camera lens. The only autofocus system that does rely on picture data is the type that measures contrast in the image and adjusts the focus until the maximum amount of contrast is achieved, but the determination of the maximum is made relative to other focus settings.
    Quote Quote  
  10. CodecVisa is a powerful real-time analyzer for H.265/HEVC, H.264/AVC/MVC, GOOGLE VP9/VP8, MPEG2 Video and YUV video codecs. It can analyze H.265/HEVC, H.264/AVC/MVC, GOOGLE VP8/VP9, MPEG2 Video on different formats.

    With lots of unique features like Decode Stage Display, CodecVisa is a perfect tool to analyze and optimize H.265/HEVC, H.264/AVC/MVC, GOOGLE VP9/VP8 and MPEG2 Video codecs, to learn and get in-depth knowledge with your eyes.
    Last edited by acheter; 29th May 2017 at 09:36.
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!