RSS FeedEarlier this year I received a case in which I was asked to “enhance the video” to determine who fired first in a shootout in a strip mall. There was no sound recorded, and there were three camera views recorded at this strip mall. Two views were of the exterior and showed one of the shooters fairly clearly, but didn’t have enough resolution to show much detail of the other. The third view was of the interior of one of the businesses in the strip mall.
Enhancing the video wouldn’t provide the detail to determine who shot first, but synchronizing the three videos did.
The video of the interior of the business showed several people sitting near the window. In one frame they are sitting, in the next they are jumping away from the window and seeking cover. By synchronizing this video with those from the exterior, one can see a movement consistent with recoil in the frame prior to to the people jumping.
In this case, it wasn’t enhancement that was needed, but synchronization of multiple videos.
G
Comment by Howard Patterson — October 27, 2009 @ 3:08 pm
Great to see you posting, George. Don’t see how you can find the time. Were all the videos from the same system in the mall? Or from different systems?
Comment by admin — October 27, 2009 @ 6:45 pm
Thanks Howard. Yes, all the videos were from the same system, which is what enabled them to be synchronized. If they were from different systems, there may have been other visual clues to help synchronize them, but in this case I didn’t need to rely on that.
Comment by Bob McMicken — October 28, 2009 @ 11:51 pm
George…
How about a blog focusing on digital camera resolution. The Costco in Colorado Springs just added an Epson 7800 printer and I’m getting some wonderful 20 x 30 enlargements from my six megapixel Nikon D70 and my wife’s five megapixel Fuji point and shoot. I can’t see much justification for anything over today’s twelve megapixel Nikons unless you’re doing landscapes or fine art still lifes, which I ain’t.
Likewise, your blog on the low rez video commonly obtained from CCTV surveillance systems struck a cord, since I designed and installed fifty-odd systems in various Macy’s stores back in the day. None were hi-rez because the technology did not exist in fifteen or twenty years ago, at least not at a reasonable price.
When asked about camcorders and electronic projectors, I’ve been telling folks to invest in equipment that will go up to 1080i because, with so many folks having home theater systems, that kind of resolution will be expected by jurors and attorneys.
Also, the synchronization issue surfaced another pet peeve of mine -failing to set the camera’s clock with either a time hack from dispatch or a cell phone, and failing to synch the clocks on all cameras when more than one photographer is shooting a given scene.
Added your blog to my favorites page and, if I can find a wi-fi connection in the conference center, I’ll show it to the locals in Longview.
Bob
Comment by stephen appelbaum — October 31, 2009 @ 12:12 am
Hey George:
Great idea…this blog.
You hit on another issue with requested for “enhancement”…these people are watching too many tv shows. They dont realize the movies start with a sharp photo and blur it for effect, making the magic of enhancement seem as easy as hitting a few keys on the computer. We call it image processing here…I hate the word enhancement.
Also, you point out another issue. They ask for enhancement because they dont know what else to ask for. They bring us hoofbeats, and want horses, when maybe they really need zebras. That is why I NEVER quote or commit to any project until I see the material.
The pet peeve I have is the placement of security cameras…up high and far away! Why cant banks put cameras at eye level on the counters, hidden in some part of the fixture, so we are not dealing with images that take up 1% of the image area??? One bank I deal with (remaining nameless)seems to have NO cameras at all. Since I knew that was not possible, I spent time one day (while in line) searching for them. I found little black dots embedded in the housing of the light fixtures aimed at just the right spots. When I told the manager, who knows what I do, that I “found” them he discreetly showed me the others….very cleverly built into the decor. He also told me that the decal with the bank name on the door is exactly five feet high…for obvious reasons. Nice job! But they still need cameras at eye level closer to the tellers.
Keep up the good work.
Comment by David "Ski" Witzke — November 2, 2009 @ 12:10 pm
Bob, I am not sure what you are capturing with your Nikon D70, but it can’t be used to capture palm prints, foot wear, tire tread impressions. These applications require more than a 12 megapixel camera. For example, to capture a palm print with a 12 MP camera, you could only capture an area that is 4 inches by 3 inches and maintain the required 1000 PPI. Most full palms cover an area that is significantly larger.
When capturing footwear impressions or tire tread impressions, capturing an area that is too large using just a 12 megapixel digital camera causes lose of detail and sharpness in the digital image, and creates other problems, such as lens (barrel) distortion and moiré patterns. The recommended goal is to capture footwear and tire tread impressions with a resolution of at least 500 PPI. Tire tread impressions should be captured as “segments”, allowing at least a 1- to 2-inch overlap in each segment.
Again to achieve this goal, you have to have a 12 plus megapixel digital camera. For example, the Canon EOS 7D, with a resolution of 5184 x 3456, would allow you to capture an area that is approximately 10.5 inches long by 7 inches wide while maintaining a resolution of 500 pixels per inch. While this camera has a high resolution, it also has a small imaging sensor (22.3 mm x 14.9 mm), which is less than a standard 35 mm frame size (36 mm x 24 mm). The small imaging sensor can cause distortion in the digital image because of its 1.6 focal length multiplier. The Canon EOS 5D Mark II digital camera is a 21.0 megapixel camera with a resolution of 5616 x 3744 using a standard 36 x 24 mm imaging sensor, which provides a focal length multiplier of 1. This means that you can capture an area that is approximately 11.25 inches wide by 7.5 inches without lens distortion.
The Epson 7800 printer is great for printing high resolution footwear and long tire tread impressions! You can also print large (long) objects in your office if you have the Epson Stylus Photo R1900 (cost is approx $400 and you can print images up to 13 inches wide by 44 inches long with the Epson roll paper).
Ski
Comment by Steve Everist — November 3, 2009 @ 8:04 am
Ski, who requires 1000ppi? This is something that continually confuses me. Although I often reference it for the same purposes you have – showing how much area a specific camera can cover at this resolution, I always throw in the fact that this ‘minimum’ resolution has no basis as to why it’s a minimum. The only thing I can think of is that it’s somehow been driven by a combination of vendors and standards organizations so that they can talk to each other.
But as someone who does comparisons, I find it to be potentially detrimental to casework. In a workshop I’ve given earlier this year, I purposely have an image of a latent print that I shot with my old Nikon 995 that calibrates out to under this level. I’ve never had one examiner say that the resolution was too low for comparison purposes. When shooting palms (I won’t speak to tire/tread as I don’t do these comparisons), one can simply overall the large area and then do subsequent closeup areas of detail that can be captured at a resolution that will sufficient for comparison. If an image calibrated to 99X, would it have to be ignored as marked not of resolution value?
If I remember correctly, the Hayden case used the Kodak DCS420 with a sensor of 1524X1012, and it included palms. There’s no way it could even come close to calibrating near 1000ppi. Unless the images I have aren’t original, one of them doesn’t even calibrate to 400ppi – and it is often referenced digital imaging case law.
To me, saying that 1000ppi is a required minimum resolution is like saying that X number of Level 2 details (or points) is the minimum necessary to conclude an ID of a print – it may be used as an administrative standard, but has no scientific basis.
All that being said – I usually flatbed scan at 2400ppi by default – I’m a Resolutionist…
Comment by Erik Berg — January 29, 2010 @ 8:37 am
Steve,
To answer your question, the “standard” came from Steve Meager at the FBI via the NIST conference in 1999 I believe. The FBI and SWGFAST wanted 1200 ppi as a minimum. I pointed out that the resolution of several images in Hayden were only 289 ppi. The UK presented test results from 34 Latent Examiners that showed a resolution of 600 ppi actually produced one additional identification over the same set of images captured at 1200 ppi. NIST compromised by establishing a standard of 1000 ppi for the purpose of searching automated fingerprint systems. The UK established a standard of 600 ppi because they weren’t willing to pay for the additional storage and bandwidth 1000 ppi would require.
So, that’s where the 1000 ppi standard came from.
Erik
Trackback by home security systems cctv — August 14, 2011 @ 8:14 pm
home security systems cctv…
Nice site, I think you like this latest post on home security systems cctv…
Comment by Xehqzvfl — December 21, 2011 @ 9:06 pm
comment3, walgreens viagra price, 347,