I understand that Photoshop now has a tool that allows you to calibrate an image that does not contain a scale. Does anyone know what version that tool showed up in and how you do the calibration?
Thank you!
Calibrating an Image Without a Scale
-
josher89
- Posts: 509
- Joined: Mon Aug 21, 2006 10:32 pm
- Location: NE USA
Re: Calibrating an Image Without a Scale
I wasn't aware of this tool and if it exists, I wouldn't know how it would work unless you have something in the image that you have a known size.
There was a lecture at the IAI in Reno where someone presented on how to use objects within an image to assist in calibration for AFIS submission. I'll see if I don't have the PPT of it.
Otherwise, Steve E. might be the best to answer this or The Great George Reis!
There was a lecture at the IAI in Reno where someone presented on how to use objects within an image to assist in calibration for AFIS submission. I'll see if I don't have the PPT of it.
Otherwise, Steve E. might be the best to answer this or The Great George Reis!
"...he wrapped himself in quotations—as a beggar would enfold himself in the purple of emperors." - R. Kipling, 1893
-
mgirard
- Posts: 24
- Joined: Fri Apr 20, 2007 4:19 pm
Re: Calibrating an Image Without a Scale
Not sure if this would help in your scenario.
I recently had a case where the investigating agency submitted images of multiple finger impressions and a palm impression with no scale and no other objects to use as a scale. The palm contained 3 looping formations in the hypothenar (I've never seen this previously) and was suitable to search in ABIS; however, I needed to be able to calibrate the image. I based the calibration on an average number of ridges per cm. I used the estimate of 20 ridges per cm based on other cases I was working at the time, my own palm, and information in Quantitave-Qualitative Friction Ridge Analysis, with the understanding that ABIS has some tolerance if the calibration is slightly off. I searched it and was able to get an NGI hit!
I recently had a case where the investigating agency submitted images of multiple finger impressions and a palm impression with no scale and no other objects to use as a scale. The palm contained 3 looping formations in the hypothenar (I've never seen this previously) and was suitable to search in ABIS; however, I needed to be able to calibrate the image. I based the calibration on an average number of ridges per cm. I used the estimate of 20 ridges per cm based on other cases I was working at the time, my own palm, and information in Quantitave-Qualitative Friction Ridge Analysis, with the understanding that ABIS has some tolerance if the calibration is slightly off. I searched it and was able to get an NGI hit!
-
Boyd Baumgartner
- Posts: 567
- Joined: Sat Aug 06, 2005 11:03 am
Re: Calibrating an Image Without a Scale
I'm not sure Adobe has such a feature, I scoured the internet. Calibration by definition is the number of pixels per unit of measurement. Counting pixels is easy. It's the unit of measurement that needs to be provided, hence the scale.
That being said, I agree with mgirard's general approach.
Here's another one. Use the Mean hand data from this study https://www.sciencedirect.com/science/a ... 0025#t0010 and set the calibration manually using the mean width of the hand as the ruler. This came from a thread on Sandy's mailing list regarding whether or not an adult hand would hit on a registration that occurred when the person was a child in AFIS. The consensus was that AFIS had an approximately 25% tolerance. The thread is from 5/22/19 if you want to look it up.
That being said, I agree with mgirard's general approach.
Here's another one. Use the Mean hand data from this study https://www.sciencedirect.com/science/a ... 0025#t0010 and set the calibration manually using the mean width of the hand as the ruler. This came from a thread on Sandy's mailing list regarding whether or not an adult hand would hit on a registration that occurred when the person was a child in AFIS. The consensus was that AFIS had an approximately 25% tolerance. The thread is from 5/22/19 if you want to look it up.
-
josher89
- Posts: 509
- Joined: Mon Aug 21, 2006 10:32 pm
- Location: NE USA
Re: Calibrating an Image Without a Scale
In preparing for our first ANAB audit, I didn't want to calculate measurement uncertainty and use that because I felt it was immaterial to how AFIS operates as the only measurement device we use is a ruler for latent print photography - and we don't report those measurements anyway.
So, I scanned in my print with a scale and told AFIS a centimeter was a variety of millimeters. My print hit when AFIS thought a centimeter was anywhere from 7mm - 15mm. That's closer to 30% variance. When calibrating an image we might be off by a few pixels, but shouldn't be off my more than one millimeter so I showed that measurement uncertainty wasn't necessary for accreditation.
At least since BIS 4 for Morpho systems, they have a calibration tool that uses the number of ridges as a quick and dirty way to size a print properly before searching - I haven't used it outside of training but it worked then. I'm sure this was based on the research available that says the average number of ridges per unit of measure.
So, I scanned in my print with a scale and told AFIS a centimeter was a variety of millimeters. My print hit when AFIS thought a centimeter was anywhere from 7mm - 15mm. That's closer to 30% variance. When calibrating an image we might be off by a few pixels, but shouldn't be off my more than one millimeter so I showed that measurement uncertainty wasn't necessary for accreditation.
At least since BIS 4 for Morpho systems, they have a calibration tool that uses the number of ridges as a quick and dirty way to size a print properly before searching - I haven't used it outside of training but it worked then. I'm sure this was based on the research available that says the average number of ridges per unit of measure.
"...he wrapped himself in quotations—as a beggar would enfold himself in the purple of emperors." - R. Kipling, 1893
-
Steve Everist
- Site Admin
- Posts: 551
- Joined: Sun Jul 03, 2005 4:27 pm
- Location: Bellevue, WA
Re: Calibrating an Image Without a Scale
Like Boyd mentioned, it's pixels per unit of measurement. You need to be able to find the distance of something that's in the image. Sometimes this isn't able to be done (piece of non-ruled paper for example). We've had a couple of them that I've helped with. One was a water bottle with a lid. We were able to find the same type of lid and measure that. Once we knew that size, we could do the math to calibrate the image.josher89 wrote: ↑Thu Aug 27, 2020 5:23 am I wasn't aware of this tool and if it exists, I wouldn't know how it would work unless you have something in the image that you have a known size.
There was a lecture at the IAI in Reno where someone presented on how to use objects within an image to assist in calibration for AFIS submission. I'll see if I don't have the PPT of it.
Otherwise, Steve E. might be the best to answer this or The Great George Reis!
Another one was on a handgun. There were a couple of things in the handgun that we could get the distance from. Using other photos of the same handgun, that did have a scale, we could calibrate the distance between the two points. Then we applied that to the image with the print to be searched.
At the IAI conference in Sacramento, Tim Fayle gave a Trending Topics short lecture on how they did it in a case where the suspect was holding a pill in their hand. The detail was good enough to search the palm. He found the dimensions of the specific drug that was in the image and used that size to calibrate and do the palm search.
Steve E.