businesstraveller2

Is it just me, or did anyone else realize that the nudeoscope is not addressing the correct problem? Nude people are not the threat it's what's between the outer clothes and the skin. So why are they showing mostly nude pics in the images? I'd be curious to hear what others know about the imaging side of this technology - not the sensor/data collect but how they process the data to produce the picture. My sense is that companies who produce these are heavy on hardware and light on software. They should be producing pics of what's between the outer clothes and the skin and no need to show all that skin because it's not a threat AFAIK. I'd like to be enlightened by others on the forum who may know more....

janetdoe

They are doing most of this by using hardware and exploiting the physics of low-energy x-rays, and probably very little of it in software.

They have identified a narrow window of x-ray energy where human flesh gives a bright signal compared to most other materials. They detect the flesh, and everything that blocks the flesh signal is an 'anomaly'.

Clothing is usually thin and not very dense, so most of the x-rays just pass through the clothing, slightly attenuated, but fairly uniform. That is why bulky clothes or skirts that drape around the legs generate secondary pat-downs - they either attenuate the signal too much or unevenly. Otherwise, the clothes probably just don't produce much of a signal.

The x-ray detector is probably at a fixed angle from the beam and may have some filters or slits to improve the signal-to-noise ratio. They probably raster the beam and detector as a single unit. They read the voltage signal from the detector for a certain position, and that is the intensity value of that pixel.

This generates the basic image, very little software or image processing involved.

At this point, software may scale the brightness of the image or increase the contrast. Maybe they use a standard image sharpening algorithm. They detect the head (the small round blob at the top of the image) and blur the center of it. It looks like some software versions have an edge-detection or feature-detection algorithm that draws yellow boxes around areas of interest.

The rest is probably training TSOs how to look at the image and associate certain contrast patterns with certain threats. Or at least to tell when something 'looks funny'. Eventually they should be able to develop software to recognize patterns faster and more accurately than TSOs

But they can't really eliminate the flesh - it provides the 'light' and they look for shadows blocking the light.

businesstraveller2

Thanks for the more detailed info. The rendering could be much improved I think - eliminating passengers concerns about people seeing them nude - but it requires some work and possibly another set of cheap sensors. I will think about this. I work on imaging and data fusion algorithms for my research and I find this problem very interesting - at the very least because I travel often and don't want to go into those machines.

DevilDog438

Yes, I am being tongue in cheek here, but I believe the portion I bolded is a real goal of the TSA.

N965VJ

The various Nude-O-Scope manufacturers have been experimenting with a software upgrade known as Advanced Threat Detection, but the TSA has not approved it.