[Leaplist] Linux configurable ip surveillance camera systems

Richard F. Ostrow Jr. rich at warfaresdl.com
Tue Feb 23 14:04:33 EST 2010


I had considered this a few times in the past... I'd come up with the
following requirements:

(1) cameras _must_ be IP, and must have either an API/complete protocol
documentation (I've seen this on a few camera models), or unix drivers
that can tweak every knob the thing has.

(2) To be useful, they really need to be night-vision capable... these are
common enough, though

(3) The system I've been contemplating creating is one where I have
several cameras set up around the house, all of which record to a large
ring buffer in the house somewhere on a fault-tolerant machine (preferably
one that it takes some time to get at and do anything to, so that anyone
doing anything with it is caught on camera). Once per frequent interval
(defined as whatever I feel I can afford/is reasonable), it should send a
high quality frame from each camera over the net to a much smaller ring
buffer to a remote machine - in case the security system itself is swiped
:). In the event something bad happened, you now have video surveilance of
what happened... even if the system itself was swiped. Of course, I'd also
take this a step further and encrypt the remote ring buffer for privacy
reasons, but you get the idea...

Beyond that, you can also fiddle around with motion triggers and
similar... but I consider the above to be the bare minimum that I'd be
willing to work with.
On Tue, February 23, 2010 12:08 am, Kevin Korb wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> I will agree with most if not all of what Fred said.
>
> Years ago my company set these up: http://www.visionquest.net/Video.php
> They are a neat marketing gimmick which is why they are publicly
> available and they allow me to quickly see what is going on from home if
> the real security system is tripped but they provide no additional
> security on their own.  In fact the quality level is so low we don't
> bother recording with them at all.
>
> Technology wise they are web cams that connect via CAT5 ethernet cable
> (but not via IP).  4 of them connect to a little black box that then
> connects to a PC via a single composite NTSC TV capture card and a
> serial (RS232 not USB) port for control.  Crappy standard definition
> NTSC divided by 4 gives you an idea of what kind of quality we would get
> if we tried to record with them.  It is slightly better than what they
> transcode for the web but it is still almost worthless.
>
> The really sad part is that the software these things came with actually
> supports motion detection and even fencing.  The problem is that the
> image quality is so grainy that it constantly sees full screen movement
> based on the built in interference.
>
> On 02/22/10 23:54, Fred Moore wrote:
>> Dan Cherry wrote:
>>> Anybody have any experience with ip based surveillance systems?
>>>
>>> D-Link has some reasonably priced hardware, but the initial set up is
>>> through Windows.
>>>
>>> Also, has anyone any thoughts on Linux accessible surveillance dvr's
>>> (vs. feeding IP camera output directly to a PC for storage)?
>>>
>>> I'm considering 4 cameras with a dvr expandable to 8 cameras.
>>>
>>> Finally, should I just fall back to older analog cameras that feed
>>> into a DVR that is IP accessible?  That seems like the best bang for
>>> the buck, but the least flexible.  Hmmm, come to think of it, how much
>>> flexibility do you need in a video surveillance system?
>>>
>>> Any thoughts on the subject would be welcome,
>>> Thanks,
>>> Dan
>>>
>>
>> Do that for a living..  My knowledge is yours for the asking.
>>
>> Don't think about Web cams as they are not good IP cameras they don't
>> get you to a codec till the high bandwidth video gets to the computer.
>>
>> Camera Trade offs,  CCD vs CMOS imager, and size, equate to quality,
>> light level, streaming rates, and video compression at cameras.
>> Personally I would never install a CCD imager camera, but 90% of the
>> camera's on the market use this technology, why because of the advent of
>> phones with cameras built in.  The CCD imager has forced a huge price
>> reduction.  Don't know where you live, but I can show you the
>> difference.
>>
>> DVR.  "Digital Video Recorder" (not for recording IP camera), DVR's have
>> analog video input.  Almost all DVR hardware have A/D converters on
>> input.  Mostly not good deals because of the unknowns internals (more
>> explained next).  Stay with Analog or IP, and don't mix them.
>>
>> Most  DVR, start with a 4-CIF video image (input), only to have it
>> immediately  reduced to a 1-CIF image via pixel trashing, then run it
>> through a lossy codec and store it, in MJPEG,  in MJPEG every frame
>> carries 100% of the infomation and the storage goes way up compared to
>> MPEG4, H.264 etc, when you reach these codec's you input 1/4 the image,
>> and move to P/I frames.  P frames very from recorder to recorder if we
>> assume a P frame of 1 second and are streaming at 15 FPS, recording
>> looks fine in playback but suddenly rewind viewing, and searching is
>> very ugly why because the I frames can not be reconstructed backwards,
>> so rewind is a hideously jumpy 1 FPS image.  The problem is you don't
>> know on cheap junky recorders what they are doing inside at the codec
>> level.  Take H.264 "best bandwidth compression" we have,  While the P
>> frame rate is part of the standard, some recorders are doing P frame
>> rates as far out as 15 seconds, unless they see movement and then they
>> record more P frames.  Why? they don't have the processing power, or the
>> storage capability to store it properly.  All of this costs money.  This
>> is not goodness.
>>
>> Flexibility: unless you have something with good analytics, you don't
>> have a security system, you have a recording system. Security systems
>> notify of something going wrong, or at the very least have a mechanism
>> which will flag the recording of an event.  Otherwise if you want to
>> know when something happened, start watching 24 hours/day of recordings.
>>
>>
>>
>> Analytics like, motion, virtual fences, Image identification (car,
>> person), camera moved, camera shaking, identification are all
>> availability.   The better they are the higher the costs.  Most simple
>> motion analytics at the camera fall down when place outside and the wind
>> is blowing, or it is raining.  No camera manufacture has resolved this.
>> Good VMS (Video Managements Systems) have solved this years ago, can you
>> say processing power?  The higher resolution of the camera, the more
>> processing power it takes.
>>
>> I deal with things like virtual video fencing, and image recognition
>> every day.  Its all processed at the VMS, why Camera's can distinguish
>> movement, but was it a bird, dog, person, car, rain, or wind.  Depending
>> on quality you can go from 1000 alarms/day to view or search, or 1 real
>> event per day.  Cost is the difference.
>>
>> There are a couple of Linux solutions at the VMS level, but don't
>> recommend them.  All I know of want to also be compatible with windows,
>> unix, Mac..  can you say Java.
>>
>> Some of the things to consider.
>> codec between the recorder and the camera?  codec's are the trade off
>> between  network bandwidth, resolution, storage requirements, and
>> processing power.
>>
>> D-link is total junk, it is built for 1-CIF camera's (1/4 analog
>> signals), if you are going IP you need 1D capability, and the horsepower
>> to manage it, or there is no reason to go with IP.
>>
>> This is a huge topic, or we can move it to BS as this is not Linux
>> centric.. I will assist you if you give me some more information or give
>> me a call..  cell 407-304-0709 Regards.. Fred
>>
>>
>
> - --
> ~*-,._.,-*~'`^`'~*-,._.,-*~'`^`'~*-,._.,-*~'`^`'~*-,._.,-*~'`^`'~*-,._.,-*~
> 	Kevin Korb			Phone:    (407) 252-6853
> 	Systems Administrator		Internet:
> 	FutureQuest, Inc.		Kevin at FutureQuest.net  (work)
> 	Orlando, Florida		kmk at sanitarium.net (personal)
> 	Web page:			http://www.sanitarium.net/
> 	PGP public key available on web site.
> ~*-,._.,-*~'`^`'~*-,._.,-*~'`^`'~*-,._.,-*~'`^`'~*-,._.,-*~'`^`'~*-,._.,-*~
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v2.0.14 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
>
> iEYEARECAAYFAkuDYrQACgkQVKC1jlbQAQfBVACg3X+ACHyATzcQOOKwkapemTx1
> Mb4AoPZYf1O2V/8buqUsufp8dgKcRwsQ
> =tNDu
> -----END PGP SIGNATURE-----
>
> --
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
>
> _______________________________________________
> Leaplist mailing list
> Leaplist at leap-cf.org
> http://lists.leap-cf.org/mailman/listinfo/leaplist
>


-- 
Life without passion is death in disguise


-----------------------------------------
This email was sent using SquirrelMail.
   "Webmail for nuts!"
http://squirrelmail.org/


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.



More information about the Leaplist mailing list