Firstly, I would like to thanks Pylgrym for providing me the two original (huge!) videos.
With my partners, we are actually working on the development of two new units for the analysis software that are:
- The first part of an authentication tool that is able to provide a quick assessment about the validity of the metadata associated with any JPEG
file, either in a single or batch process. The second part will be the continuation of the constitution of a camera/camcorder database with lots of
possible technical characteristics for a given camera (resolution, shutter speed, aperture, ISO, and so on...) plus the JPEG compression signature.
The finality of it would be to compare all these characteristics of a given photo to this database.
- The second unit could be a very useful tool as, nowadays, we are flooded with lots of reports that appear to be afterwards, just Chinese (or "Thai")
lanterns. This is still a work in progress, but a preliminary presentation will be available on our main
before the end of March.
Anyway, and to be short, this tool/algorithm is based on "chromaticity
", based on hundreds of
original "real" chinese lanterns photos, with calculated average chromaticity, minus the dark background noise and the saturated pixels. The same
chromaticity database is constituted as well for any kind of light source upon a black background sky (airplanes/helicopters, RC models, stars/planets
So I'll try this new tool on the first video, as the classic orange hue leads some members here to think that it could possibly be just one of these
damn chinese lanterns.
Before showing you the results, please keep in mind that this is not in any case a definite answer as to what it could be. This should be considered
just as an helpful tool for any analysis/investigator.
I've extracted three jpeg frames in hi-resolution of the video, where the light can clearly be seen, then run the soft
The little black circle materialize the chromaticity value of the object in the Maxwell's
, with the red polygon being the actual possible values for the lanterns.
You'll notice that, in all the three examples above, the assessment is "likely" but, sometimes, the value falls in two or three polygons, thus the
"possibly" assessment. If it's outside all the possible polygons, then the assessment will be "unlikely".
Unfortunately, there's not enough visible reference points in the video to try to calculate size/distance or speed of the object, so I guess that the
final conclusion will be (for me, as a photo analyst) "likely a chinese lantern".
I'll come back in this thread in few days to comment the other video as well.
edit on 15-3-2014 by elevenaugust because: spelling