Counter Terror Gazette

X-ray Image Enhancement: optimising the viewpoint

14 Jun 2011

X-ray Image Enhancement: optimising the viewpoint

From zoom to negative and outline tracing to high penetration, the X-ray operator has numerous image enhancement options at his or her disposal. How can they be turned from gimmicks into useful interpretation tools?  Paul Quellin describes and illustrates just what can be achieved with tried and tested X-ray technology.

Although it had been around for quite some time, X-ray screening really became firmly established as the standard for passenger cabin baggage inspection in the early 1980s. Operator consoles were relatively simple and there were few image enhancements available; indeed with some early machines which were still around in the early eighties, there were no enhancements at all.  I can recall using the one remaining fluoroscope at Manchester Airport, shortly before it was hastily spirited away when it was decided it probably wasn’t safe for us to operate!  We have come quite a long way since then, but we might ask if we have come far enough given that we have had three decades to improve. The basic principles of X-ray – laws of physics – can’t really be changed. The quality of the image presented to the operator has certainly improved, and there have been some significant advances, such as the development of multi energy colour and dual or multi view systems. Latterly, it really seems to be the operator workstations and their image enhancement functions which distinguish the main machine manufacturers from each other.

So which enhancement functions are actually useful? We need to consider the environment first.  We still tend to take a generic approach to enhancement functions; it would be all too easy to write only about passenger cabin baggage screening, when in fact the needs for a cargo screener are now significantly different. It is partly a predictable progression and partly through restrictions that the contents of passenger cabin bags have changed somewhat over recent years.  Presently, electronic items are generally easier to screen than they were just ten years ago, whereas some cargo screeners are still dealing with large and dense items every time they operate the equipment. To help us decide what might be useful, I have put image enhancement functions into three broad groups:

1: Contrast     2: Colour     3: Magnification    4: Others.

Contrast Functions

These functions are mostly about ability to provide what I call “Apparent penetration” for X-ray opaque areas, though some of the functions will help provide more apparent detail in lower density areas.  Of course we should be clear that nobody has a machine in which the operator has any control of the X-ray tube voltage and no operator can decide to increase the X-ray power to see more.  Operators can only ever change contrast settings on an image which is already processed.   These functions generally involve contrast and gamma settings and can lighten darker areas in an image in such a way as to provide some detail in areas where it may have seemed opaque (where the X-ray beam was unable to penetrate).  This is illustrated with two images of a piece of lead sheet shielding and a small semi automatic pistol (figures 1 & 2).

Of course in the cabin baggage scenario, this is not a realistic illustration, as any X-ray opaque area of this size should cause the operator to reject the item for a thorough hand search.  My own experience suggests that cabin baggage screeners are using “High Penetration” type functions less frequently, as they just don’t have the need as frequently, unless they are at an airport near a lead crystal factory outlet (though of course the same rule about rejecting things for search should still apply) generating an unusually large number of dense objects in passenger’s carry-on items.

For cargo and mail screening however, these functions are important and are likely to be needed frequently.  A point worth making at this stage is that these functions only provide apparent penetration, they cannot reveal detail which was never there in the first place; they can only make it easier for the human eye to detect.  The opposite of the penetration effect is to darken the image, which tends to make details in the lower density organic materials a little easier to see.  There are likely to be fewer occasions when this may be called for in aviation security, and it is likely to be used where we already have some concerns that something organic is concealed in the bag or item.  In this case the plastic knife concealed in the shoe on the left probably shows equally well at either setting, as illustrated in figures 3 and 4.

Making these functions available in gradual increments as opposed to moving straight to one fixed setting is another way of providing more options.  Some manufacturers have also made efforts to combine the effects of lighten and darken functions.  This provides apparent penetration in denser objects whilst enhancing detail in lower density areas.  This function generally works quite well and does prove popular with some operators.  Some Rapiscan machines have been configured to display the image in their “Crystal Clear” mode as the default whenever the conveyor is stopped and I suspect this probably helps slightly with reject rates, though I can’t confirm that.

Other manufacturers have similar options and these too tend to provide a sharper look to the image.

Colour Functions

Probably the easiest colour function to analyse is no colour at all.  Monochrome or black and white images may still have a place and though not perfect, the human eye is pretty good at working with grey shades.  All well and good, but then we usually show the black and white image on the same monitor; a monitor designed to present colour images.  Nonetheless there are still some situations where the operator may perceive that fine detail is clearer in black and white than it is in colour and if the operator is happy, then there has to be value in that.

Next we have inverse or negative imaging. These are exactly as the name suggests; you can see a negative image i.e. dark becomes light and light becomes dark, as illustrated by figure 6.  Some will work in both colour and black and white.  There is an understanding that fine wires may be easier to trace where they are displayed as white against a black background.  I can’t really say that I have any proof of this and I can’t point to any devices we have built where this makes a difference in terms of detection, as the screeners have usually spotted another one of the components before getting to this stage and we would always want them to look for the explosive and detonator (blasting cap) first. This function is probably worth keeping; though don’t expect to see it being used very often by most operators.

There are other colour functions around and I can’t help but feel some of them are a waste of workstation space and operator time.  Take for example organic and inorganic only functions, sometimes referred to as stripping.  First of all let’s deal with the names.  Organic only or inorganic stripping  seems to imply that the machine can strip away denser materials leaving the low density threats we were interested in, in all their orange glory.  Well, we are back to the fundamental limitations of the technique here.  If the X-ray beam was blocked by something, then detail can’t be put back in by some post processing enhancement.  Of all the IED training aids I have ever built, I only ever had one example of an explosive concealed inside a denser material, appearing more clearly when an organic only function was applied.  Again, the information was actually present in the original picture; it just looked a little more obvious with the function applied.  As far as I can ascertain, organic and inorganic only type functions really remove the other colours from the image and if something had to go from the workstation, this is where I would start.

Now if removing one or more of the colours from the image is not useful, then I wonder about the wisdom of having this function available in gradual increments.  Operator personal preference is an important consideration though and if someone believes there is a worthwhile use of these functions, I would be glad to listen.  I think these options are summed up by one manufacturer’s brief description in the pages of their manual:

“Image display mode Organic Only* (OxO =O²)…..” Well, that’s helpful!

Pseudo Colour

I don’t want to spend long on this, as it doesn’t really merit much attention these days.  All machines were black and white when I first used them; I was there when pseudo colour was introduced and at the time it sort of had a place, but it is time to move on. Anyway, could you look at colour combinations like in figure 9 for 20 minutes at a time?

MORE ONLINE