Any ideas on how to do it?
heres a pic from the video clip i got...survelliance camera
here's specs on camera if it'll help out any
•1/3” Color Sony CCD,
•480TVL, 0.4 lux
•70° field of view (3.6mm lens)
•Pick up Element: 1/3" Color CCD Image Sensor
•Number of Pixel: 512(H) x 492(V)
•Min. Illumination: 0.4 Lux / F2.0
•S/N Ratio: More than 48db (AGC Off)
+ Reply to Thread
Results 1 to 19 of 19
If that picture is the clearest frame, then it's impossible. There just isn't enough information to pick out any distinctive shapes. Sometimes different frames can reveal different unique characteristics and by piecing them together you can reconstruct the image, but this image is out of focus and overexposed. There is no identifiable shape in that plate. You can't even tell how many characters it's supposed have, much less what they are. No amount of enhancing can bring out something that isn't there.
Originally Posted by johnconnerOriginally a member since 2001, LONG LIVE TARAN's!!!
I'm Johnconner leader of the ?.>/,./
OH wait wrong forum.....
Yeah I figured it was a longshot so what about camera 2?
Man what the hell remember that movie with kevin costner what was it no way out? They were using computers with the power of a casio watch to clear up that picture remember
The ones I've helped nab have usually been from recognizing the vehicle later in the same general area. Canvassing the neighborhood with a photo helps sometimes if anyone was paying attention. Then I would turn the investigation over to the DA and let them follow it up. A few times I've been lucky enough to capture a face and that was a positive ID.
Never have been able to get a license plate clearly enough to be of much help. That takes a good quality camera and a decent recording format and a lot of luck. Most regular outdoor surveillance cameras end up with dirt on the lens or spiderwebs and are recorded on older timelaspe VCRs, so not much to work with at times.
I did manage to upgrade our equipment and get some higher definition cameras (600 line) and a couple of C-mount telephoto lenses and a low light (.0001 lux) camera for night use. I also picked up a DVR with H.264 encoding and motion detection and two 500GB HDDs, and it could record at 15fps or 30fps and still have room left on the drives even after a day or two. With motion detection or scheduling, a week or two.
Our Sheriffs electronic crime unit was using a low end Dazzle box for capture on a older Emachine. They apparently never watched CSI. I did introduce them to VirtualDub and Gimp.
I enhance the capped images with Photoshop and give them that, plus caps of the original for evidence, and they've done fairly well prosecuting after the DA finishes their investigation.
One place for equipment: http://www.supercircuits.com/
Dude, did you blur out the bottom of that 2nd screenshot? 'Cuz it's been blurred to a rediculous amount (after about 2/3 of the way down).
I can see the very top of the head/tail lights and car model insignia/logo, and then, WHAM, it's like a smudge/wipe.
Note: as mentioned above, you could use a video system with motion tracking (like "anti-shake"), combined with frame summation--IF you have more shots before and after in the sequence and they're not all crap either. But don't hold your breath..
yeah, I agree with redwudz. You can do some superb things with decent source footage, but the worse it gets the less you can do of use (and often, the LESS processing you should do or you'll be creating weird artifacts). This stuff is 'pert near bottom-of-the-barrel when it comes to IDENTIFICATION (it's fine for recognition).
That first pic at the top I zoomed in with Sony Vegas I thought maybe somehow I could do somethin with the vegas software to get the plate #. I know who this guy is that I got a problem with. Short story he tried to break down my front door and if it would have been wood and not steel it would have been busted down. I got a criminal mischief case against him. Misdeameanor. So if I ever catch him on my property then I got a trespass case.
Yeah I got the survelliance system pretty much figured out. I've got everything hardwired and I'm using a computer as my dvr. I've got my cameras power plugged into one box which uses only one outlet so that was pretty cool. I need to add a couple of ups boxes for the computer and camera power supply. Surprisingly the computer dvr (vista) works pretty good without freezing or rebooting. Yeah I still need to add a couple of more cameras. Thats the good thing bout havin more than one different angles.
This whole situation has been pretty expensive between the firearms,ammo,target range(ranch land), gas to drive to range, beer to drink at ranch, and cameras and accessories.
So is the bottom line that if I can't see the image in my recording theres no way to enhance it to make out a face or license plate?
What you need is a good high resolution still camera with a motion sensing trigger. Or maybe a good high definition camcorder. Those might give you enough to work with, especially if they are zoomed into a narrow area.
So, if I get this right, you're saying these screenshots are processed and not sources?
If you want better suggestions here, you're gonna have to give us originals to work with...
edit: and if you want any more help from me, lose the ethnic remarks! I'm married to a wonderful hispanic-american and lived in Austin for many years, so I know the people there (of various ethnicities)--that kind of crap is totally unnecessary.
Originally Posted by johnconner
It is called a MOVIE. It is fiction, you knowViva Linux!
Let me just stick to my original question I was just trying to give a little bit of the story cause in a way its kind of funny. Plus maybe someone went thru the same dill with a neighbor and went the cctv route and had some good info like redwudz gave me.
Yes those are screenshots from avi files from my computer dvr.
I guess just to wrap it up to enhance the video just use gimp and photoshop?
The dvr is recording in 640 x 480 and I have the De-inerlace unchecked and record quality is set to highest.
In your second sample image:
everything outside the highlighted area is wasted pixels. The area interest is roughly 320x200, 64,000 pixels. The license plate covers about 14x8 pixels. If you had used an 8 megapixel camera and zoomed into that small area you would have 8 million pixels (~3600x2400) to work with instead of 64,000. The license plate would cover about 140x80 pixels. You probably would have been able to read the license plate easily.
A clear license plate at ~140x80 (a digital camera image wouldn't be this sharp):
And at 14x8 (again, a digital camera will not give an image even this sharp):
You will never be able to read the text in the smaller image.
I'll do it....I'll pawn my 52 in tv and my truck...
Set up a motion-sensitive pencil cam in the back of the p/u (or even inside the cab pointing out the back)