How can I infer parameters such as distance from metadata?

Question

Is it possible and how can I infer a parameter like the distance or the illumination from the chart, by looking at the metadata of the picture ?

To give some context, I am working on a chart with smartphones, a chart mainly looking like the e-sfr ones. I have aruco markers all around which I know are exactly 1cm a side. I also have MTF plots for this chart.

Looking at EXIF data I have lots of information like F-stop, exposure time, ISO, focal length, ... Is it possible to compute external parameters from these ?

Answer 1

Since you know the size of the aruco markers, the focal length of the lens, and the sensor size (or the effective focal length in 35mm terms), you could use the recorded size of the markers w/in the image to determine the recorded FOV, and therefore the distance at which they were recorded.

Various calculators for determining image aspects (distance): http://www.tawbaware.com/maxlyons/calc.htm

Answer 2

Some cameras record the focusing distance. But given the wide DoF of smartphone cameras I doubt this has any usable accuracy (assuming they focus at all).

For exposure, some cameras (my DSLR) output the measured EV in the EXIF while others (my phone) don't, but all output the aperture, exposure time and exposure bias, if any, from which you can recompute the EV.

Historically there is no EXIF support in the PNG format, even if some PNG metadata parallels EXIF metadata. A formal EXIF support was added to the PNG format in 2017, but support in major image apps looks currently spotty (exiftool supports it).

Support of formats other than JPEG is smartphone dependent. Raw could be possible in high-end nerdy smartphones, but I doubt you'll see a smartphone output PNG or TIFF. But a good quality (>=90) JPEG isn't that bad...

Answer 3

If you can see bokeh circles of faraway light sources, they appear like paper disks of diameter f/a (optical focal length divided by aperture number) would look when placed in the focusing distance. That gives you a rough manner of estimating the focusing distance based on visual artifacts. It requires such artifacts to be prominent, of course, so you are not likely going to get something useful from this approach unless you actually have an effective lens opening that isn't tiny compared to the subject you are focusing on.