Early Days of Forensic Image Processing
When I got started in Web and Internet programming I used to brag about being there "before there were pictures". But where Forensics and Image Processing is concerned I can also brag about "being there before Microsoft".
Everyone in the Forensic field of Evidence Examination would like to think that science resulted in what is now taken for granted. Absolute rubbish! A lot is due to some accident of fate. Nothing at all to do with a whole bunch of researchers in laboratories.
Take for example the technique of "Fuming" to raise fingerprints. A total accident. The result of the original supplies of Crazy Glue being hand packaged. Seems they were being returned due to fingerprints that appeared on the inside of the plastic containers. The latent prints of those folk that packaged the glue. Some astute evidence examiner figured it just might be a great way to raise seemingly invisible fingerprints.
Laser illumination? Another nuisance problem in another industry! Seems a copier salesman and an evidence examiner (from Canada) were chatting over the back fence. The new laser copier that was just introduced had a small problem. Fingerprints inside the machine glowed! Needless to say that was exactly what every evidence examiner was dreaming of...
My particular field, that of image processing, was strictly the result of a downturn in the economy at the time.
Forestry in British Columbia had been using a form of what we then called Image Analysis. Processing never entered into our minds as we never had any real practical desktop computers. But, by suitable analysis of aerial photographs we could pick out dead trees in a forest. Enter good old fate to this one. The forest industry took a downturn and we were stuck with a warehouse full of analysis machines. By a quirk of fate, the same fellow that was in the conversation with the laser copier technician (Brian Dalrymple) just happened to see an advertisement we were running at the time. Seemed to him that if we could pick out dead trees in a forest we should certainly be able to resolve a fingerprint in a complex background. He was 100% correct. We did that, and more...
Let's put this all into perspective as to what we had to work with!
First, the only commercial computer that was available was the Apple. Not exactly a high tech device, but it was mass produced and extremely reliable.
Our analysis system adapted itself very easy to digital interface, so the marriage was one made in heaven.
There were absolutely no digital cameras or scanners in 1983. But, we did have a very high resolution video camera system. The interface to a digital memory was in place. So, all we had to do was come up with some other bells and whistles to make it work. Digital enhancement was totally out of the picture. But, as we were working in a video world then enhancement was simple.
The key to the whole system was a technique called "density slicing". We could take an image and slice it into 256 layers based on the brightness of what was being viewed. Every image is made of a transition of shades from pure white to absolute black. These are called shades of grey, or technically "grey scales". In forestry we looked at the aerial photographs of a forest and sliced out areas of interest purely by the brightness of the trees. An infrared photograph of a forest resulted in a photograph where the water content of live trees resulted in a totally different image of a dead, dry tree. Reduce that picture to the now common "pixel" and you have a common means of measurement. Assign a value from 1 to 256 to that pixel and you can actually count trees if you know how many pixels a standard tree (or forest area) represented.
But, that was just the tip of the iceberg here. If you could capture an image digitally you could perform all sorts of miracles with it. Make a digital drawing of an X-ray and then compare it directly to a picture of a bone (or actual body) and you have a definite tool for superimposition. Make a digital mask of a repeating pattern, superimpose it on an object containing a fingerprint on that pattern. Then do some simple digital math and subtract the pattern. What's left? The fingerprint, naturally.
What all seems so simple now was totally impossible in 1983! Until we hit the scene that is...
Three individuals; Brian Vanderlinden, Les Hooten, and myself (Tom Wagner), worked almost day and night to perfect what we marketed as the "Infoscan Image Processing System" (all shown in the above clickable group photo). The above all sounds easy, but in actual fact we had to do considerable "screwdriver engineering" to make the system work. By mid 1983 we had a totally functional, marketable Image Processing System. A hybridization of Digital and Video technology. All we needed to do was convince the world what we could do. A formidable task when you are located in the far north west corner of the market!
What we did was make the machine available for whoever wanted to use it. Thanks to attendance at the major conferences at the time we soon developed an interest in the Forensic Community. We can truly say that the major players beat a path to our door. I never knew what would show up in the mail. Good thing the post man never asked! I still have letters from the leading examiners of the time expressing gratitude for our assistance. Joseph V. Ambrozich, Forensic Scientist from the Joliet Il. Laboratory wrote; "I have worked in the forensic science field for the past 13 1/2 years, and I must state that after observing the capabilities of this machine I am truly impressed."
We did evidence examination, facial reconstruction, bite mark analysis and many other almost impossible tasks of the day. But, we were faced with two hurdles. The first was money. Everyone wanted it, but no one had a budget to buy it. The other was the legal system itself. How could we convince a court that we were only observing and analysing not manufacturing evidence.
The latter took an actual court case. The first case that was to be held was scheduled based on evidence developed on the machine. Myself and a few other "Professionals" were called to court in California. Our evidence was based on a bloody fingerprint on a tool box. But, at the last minute the accused took his hunting rifle into the desert and referred the case to a "higher authority". In due course we did get our day in court. The evidence never did make it to the actual trial. It eventually appeared in a civil suit for "wrongful death" brought by the family of the deceased wife (reminds me of the OJ Simpson trial). It transpired that when the wife in this matter was found dead her part of the estate went to her spouse. He committed suicide and was never convicted of any wrongdoing. So, his estate went to the heirs of his will. A pretty sorry miscarriage of justice, but one that can easily happen. The wife's heirs were successful, due largely to the work of Image Processing. I personally received a letter of appreciation from Sherman Block, Sheriff for the County of Los Angeles.
So, that is the story of how Image Processing was introduced to Forensics. Personally, that part of my past is one that I fondly remember, and now every day when I see the strides made I feel proud to have been a small part of it.
Back in 1964 when I actually built my first computer (an Altair with 16K of memory) I predicted that eventually everyone would have on in their home. Who would have thought back then? My kids would brag about having a computer in the basement, and not one soul believed them. I know exactly how Marconi and Bell must have felt. I am not a pioneer in any sense of the world, but am extremely grateful to have been a small player in a very big field. If nothing else, it has gotten me a permanent exclusion from jury duty. Some of the evidence we developed is still out there in unsolved homicides.
I would really like to hear from any of the folk of those days. We certainly thought we were on the "cutting edge" in those days but in fact it was more like "the blunt edge" of the technology sword.