If a prisoner is hiding in
this truck, ORNL's heartbeat detector can determine if
the escapee is present. In the future such an
"enclosed space detection system" may be used to detect
human presence in hostage situations or collapsed buildings.
ORNL's war on crime is being waged not from its ivory tower but from its CASTLE—the Center for Applied Science and Technology for Law Enforcement, which is building a strong case for crime-fighting technology. But before the program tackles a law-enforcement problem, says CASTLE director Scott McKenney, the case must meet several criteria: The problem must require the unique resources of a national laboratory or a university—CASTLE won't compete with private-sector companies. It must be solvable. And the solution must be affordable.
Some projects have yielded crucial evidence for cases that are pending in court. Others have solved cases outright. Some examples.
"Law enforcement professionals are getting smart about what the Lab can do for them," says McKenney, "and we're getting smart about what their needs are. As we better understand one another, we can really make an impact."
If a prisoner is hiding in
this truck, ORNL's heartbeat detector can determine if
the escapee is present. In the future such an
"enclosed space detection system" may be used to detect
human presence in hostage situations or collapsed buildings.
That makes sense. You might even say it's elementary.
|
"And what do you think of it all, Watson?"
asked Sherlock Holmes, leaning back in his chair.
"It seems to me to be a most dark and sinister business." "Dark enough and sinister enough."
|
As results came in from the forensics lab, Knoxville Police Department Specialist Art Bohanan felt frustration wash over him. The suspect's fingerprints were all over the inside of the car thought to have used in the abduction. But there was no trace of the little girl's prints.
It was not the first time Bohanan had run up against this particular brick wall. In a similar case several years earlier, witnesses testified that they had seen the child in the suspect's car—but no prints from the missing child could be found there. What was it about children's fingerprints and cars that conspired against Bohanan?
Checking with others in the department and then with other law enforcement agencies, Bohanan found little to go on. The Federal Bureau of Investigation (FBI) could offer nothing. Scotland Yard had no answers. The Israeli police had not observed it, either. Forensic scientists had done little research on the fingerprints of children, because children were rarely crime suspects.
| We think the kids' fingerprints are still there, but the current technologies used by police just don't detect them. |
Knoxville Police Detective Art Bohanan lifts a child's fingerprints off a car as his
granddaughters watch. He sought help from ORNL in explaining why childrens' fingerprints vanish faster than adults'. Unfortunately, more and more of
them seemed to be turning up as crime victims. The question,
Bohanan decided, cried out for an answer.
Though not a scientist by training, Bohanan devised an experiment. He had children and adults handle two cases of Coke bottles. One case of the bottles he placed in his cool basement as the control group; the other he put in the back seat of his police car to simulate realistic field conditions. Each day for the next month, Bohanan removed a bottle from both cases and dusted them for fingerprints. He found what he had suspected all along: while the adults' prints remained, the children's prints began disappearing almost immediately; soon they were gone altogether. He repeated the experiment for an entire year, and although the children's prints lasted longer, they still vanished before those of the adults. Now that he had proof of what was happening, he turned to science—he turned to ORNL—for help understanding why.
| To tackle the problem from multiple angles, ORNL Director Alvin Trivelpiece assembled a group of scientists from a variety of disciplines. |
Knoxville Police Detective Art Bohanan and ORNL's Michelle Buchanan examine
fingerprints. Buchanan's research suggests that children's fingerprints don't last as long as adult
fingerprints because of a difference in chemical composition.
In that instant, the investigation into the girl's murder changed dimensions. It moved out of the typical police crime lab, with limited scientific equipment, and into the national laboratory, where researchers work on the cutting edge of science.
Though beyond the reach of a typical police lab, this experiment, too, began simply enough. "We took small vials and put a couple of milliliters of rubbing alcohol from a drug store in each one," Buchanan explains. "We had a group of adults and the children shake the vials between their fingers. The alcohol extracted a small amount of the chemicals off the surface of their skin."
Buchanan and her colleague Keiji Asano now had the samples they needed for GCMS analysis. Working with the ORNL chemists was a group of undergraduate students, who participated in the project as part of the Science and Engineering Research Semester program.
Over the past few decades, gas chromatographymass spectrometry has become the workhorse of the analytical chemist's laboratory, examining unknown compounds to extract precise information about their makeup. The students injected samples of the dissolved chemical compounds into the GCMS. After the chemicals vaporized there, they were separated on a heated capillary column by clinging to the viscous coating on the narrow column. Lighter, more volatile elements came through first; heavier elements held on longer.
By combining gas chromatography with mass spectrometry, the chemists learned which compounds were present. By hammering the molecules of each substance in the sample with electrons, a positive charge was given to each molecule. During this ionization process, most of the ions fell apart to form a number of fragment ions. The ions were then sorted in the mass analyzer, producing a spectrum that identified each compound, much like a fingerprint identifies an individual.
When the researchers looked at the printouts, they were amazed. The data from the experiment could be laid down in two neat piles: children in one, adults in the other. The children's prints had far more of the compounds known as fatty acids. The adult prints also contained fatty acids but at much lower levels. The adult prints, however, were observed to contain larger quantities of fatty acid esters, which are less volatile than the fatty acids that are predominant in children's fingerprints. The difference in the fingerprints was like the difference between footprints of gasoline and those left by motor oil being tracked across the floor at a service station: although chemically similar, one evaporates in moments, the other hardly at all.
Now ORNL researchers could explain Bohanan's observation. In a hot car, the lighter fatty acids in children's fingerprints were volatilizing—just going away. When police dusted for prints with powder, there was nothing left for it to stick to. From the perspective of pure science, they had answered the question. But from the perspective of applied science—the kind of science that saves lives—they had raised more questions than they'd answered.
| ORNL researchers are developing new chemical markers that make invisible prints fluoresce under hand-held lights. |
"We think the kids' fingerprints are still there, but the current technologies used by police just don't detect them," says Buchanan. Based on the results from these preliminary studies, organic chemists at ORNL are now trying to develop new chemical "markers" that police can use to make invisible prints fluoresce under special hand-held lights.
The search for an identifying compound has opened a second door. The researchers have detected a wide variety of substances in prints, including cholesterol and nicotine. They are currently investigating the potential of identifying trace components in fingerprints that can distinguish among individuals. Says Buchanan, "This raises the possibility that components in fingerprints could yield a profile of the suspect like 'female, smoker, diabetic, cocaine user.' " Soon, these tools may help police and the FBI unlock vast amounts of evidence now hidden at the scene of a crime—and help Bohanan and his colleagues bring killers to justice.
Police officers run the risk of being such targets at any time. To lower their risk of death, many officers wear protective vests made of very strong, specialty materials such as Kevlar or Spectrashield synthetic fibers, which are stronger than steel. But the same vest that keeps out bullets keeps in heat: During warm weather, police find them almost unbearably hot. And a "bulletproof vest" stops nothing if it's locked in the trunk of the patrol car.
To make this soft body armor bearable—and wearable, Moshe Siman-Tov of ORNL's Engineering Technology Division and a team of engineers are resorting to one of the oldest laws on the books: the second law of thermodynamics. To grasp one form of this law in action, stir a cup of tea with a silver spoon. Too hot to handle? That's because heat is drawn from the tea and is conducted up the cooler handle through the silver molecules, which vibrate against each other and, in turn, against your fingers. Energy moves from hot to cold. Always. It's the Law.
Siman-Tov's team is using The Law to help officers keep their cool. Call it scientific law meets criminal law.
The team brought impressive credentials to the task. Previously they had developed ways to cool super-hot orbiting lasers for the Defense Department's "Star Wars" program; they also had devised better strategies for cooling nuclear reactors that operate at temperatures above 1000°F. Cooling a human, which operates at a meager 98.6°F, should be easy enough, they figured.
Under a 1995 cooperative research and development agreement with Safariland, a leading U.S. and worldwide manufacturer of bulletproof vests for law enforcement, the engineers set to work. The theory was simple, but the rules—make that The Rules—were tough: must be used for hours without an external power source, must be light, must be comfortable, can't restrict movement, can't compromise protection.
Some currently available designs require a powered system to bleed off the heat; they work, but they bend The Rules to the breaking point (too complex). Other designs are passive but break The Rules for other reasons (e.g., too heavy).
In collaboration with Safariland, ORNL is developing a
technique to cool personnel wearing body armor to increase their
comfort and ability to perform efficiently in hot or humid climates.
That is where new creative ideas came in to help. First, to carry away heat from the body, the team selected special fibers that conduct heat better than any other material, including copper or silver. Woven into fabric that can be tailored to be worn under the vest, the fibers will remove trapped heat if engineers can offer it a cooler destination. Enter thermal energy storage material that can grab heat and store it as latent heat (turning a solid to liquid to keep the vest wearer cool) until the police officer's duties are done for the day. Then the device can be "recharged" in the patrol car while officers cruise to their next mission.
Now, even as work continues on various parts of the concept, Siman-Tov and his team contemplate new ideas that will take advantage of nature rather than fight it. Details, however, are not available for this article because ORNL engineers follow The Rules.
The new design Siman-Tov envisions would also satisfy The Rules. It would obey The Law. And it would surely save some lives. Cool.
 |
 |
| Examples of video footage of crime scenes after video enhancement at ORNL. Above are unenhanced and enhanced images of a bank robbery in Knoxville. | |
 |
This bank robber was captured on videotape for less than one second, but ORNL video enhancement enabled identification of a suspect. |
| This ORNL video enhancement helped police find a suspect in this robbery of a Knoxville convenience store. |  |
|
"You have evidently seen more in these rooms than was visible
to me."
"No, but I fancy that I may have deduced a little more "
|
The video from the Memphis convenience store security camera recorded the crime in all its horror. The robbery and murder resembled a TV dramatization, but the victim was real. Unfortunately for police, the security camera was cheap, the light bad, the movement fast. Investigators were unable to get a good look at the killer. The tape was sent to Memphis video specialists to see if they could improve the image. They could not.
That wasn't surprising, says Teresa Subich, a graphic artist in ORNL's Instrumentation and Controls (I&C) Division. The typical security camera is doomed to take low-quality video by its bad optics, age, narrow recording bandwidth, and poor light sensitivity. And it's unlikely that all the older, low-quality cameras now in use will be replaced anytime soon. So Subich and her colleagues are bringing science into the picture.
This video footage enhanced at ORNL reveals evidence of
a murder of a convenience store owner in Chattanooga.
The muzzle flash associated with the homicide suspect
appears between the two zeros, and his foot is shown under the bottom
7. This evidence has been used in court. Identification of the
muzzle flash discredited the suspect's story to the police.
A video camera uses the same concepts Alexander Graham Bell and Thomas Edison pioneered a century ago, working with weak electrical current and magnets, needles, and wax: the vibrations and impulses of sound and light can be converted into electrical signals in a charge-coupled device. If the process is reversed, the signals are converted back into sound and light. The conversions can be done in real time, as in a telephone chat or TV broadcast, or the signals can be stored as magnetic squiggles on videotapes—the modern equivalent of scratches in wax. Now, as in the days of Edison and Bell, the quality of the recording depends on the equipment and the conditions at the time.
But sometimes recordings hold more information than standard playback equipment can reproduce. To get at this information, which is often masked by interference, ORNL specialists are enhancing videotapes using commercially available equipment (Adobe Photoshop and other software on Apple Power Macintosh computers and other hardware) and their natural abilities to get the most out of these tools.
By grabbing digital information from videotapes frame by frame, then using the computer to delete interference and clear up images, the ORNL team has quickly produced a string of successes: they've helped solve a series of armed robberies, an ATM burglary, a large marijuana-growing operation, a bank robbery, and a hate crime. They've also improved the quality of video evidence from numerous other burglaries, robberies, homicides, and other crimes.
To date CASTLE's video enhancement technology has had a better than 50% success rate in providing evidence leading to an arrest or conviction in the 30 cases in which it has been used. Evidence extracted from tapes includes the identity of a person or a vehicle, which can be used in building a case for arrest and prosecution. The Tennessee Bureau of Investigation, the FBI, and police departments in Oak Ridge, Knoxville, Chattanooga, and other towns in Tennessee and in Georgia and Florida have obtained help through the program.
"We've gotten tremendous results, and demand is growing," says McKenney. "In one case, an arrest was made in one day after we got a surveillance tape and analyzed it. In another case we were able to identify the muzzle blast from a handgun used in a homicide case. The photographic evidence contradicted the defendant's testimony. A detective involved in the case said he was certain the subsequent guilty plea was a direct result of the CASTLE support."
Now, to move the technology out of ORNL and into police or private labs, Ken Tobin and Tom Karnowski, both of the I&C Division, are developing a new generation of software specifically designed to tweak security-camera videos. Such camera systems have some common problems, such as streaking caused by worn-out recording heads, blurry pictures resulting from dirty or inferior lenses, and dark images resulting from inadequate lighting. By developing computer programs that look specifically for these problems and automatically correct for them, researchers hope to help police extract better evidence from security videos at a much lower cost.
And that means the picture will be brighter for police and prosecutors.
In cases where no other identifying evidence can be found, forensic anthropologists can recreate the victim's face by modeling clay over the skull. Using pencil-eraser-sized rubber markers attached to the skull, an expert molds the clay into "average" cheeks, lips, and other features. With skill, some luck, and a victim who's close to "average," an identification can sometimes be made. Unfortunately, the odds are low and the cost is high: Only one in seven such reconstructions leads to a successful ID; on average, the work takes two months and costs $2500.
Dr. William Bass, Knoxville's well-known forensic pathologist, examines the skull of a murder
victim. Bass has been with the University of Tennessee's Anthropology Department.
Murray Marks, associate director of the Forensic Anthropology Center at the University of Tennessee at Knoxville (UTK), figured advanced technology could improve the process. Professor Marks had good reason to trust technology: his predecessor at UTK, William Bass, spent years pioneering physical and chemical techniques for identifying skeletal remains. (In the process, Bass became something of a legend, eventually inspiring Patricia Cornwell's 1994 novel The Body Farm; in fact, two characters in the book are modeled after Bass and Bohanan, and the "body farm" in the book is UTK's Anthropological Research Facility.) The collaboration Marks proposed to ORNL was not physical or chemical, but computational.
| If we had more measurements, we could do a better job of reconstructing a victim's face. One of the things that a computer can do very well is take many measurements. |
"Over the years, anthropologists have identified the thickness of tissue on a limited number of places on the skull," explains Marks. "It was clear that if we had more measurements, we could do a better job of reconstructing a face. One of the things that a computer can do very well is take many measurements."
At ORNL Marks met up with biologist Richard Mural and computational biologists Reinhold Mann and Ed Uberbacher, who were working on the problem using ORNL's supercomputers. From the outset, the researchers knew they needed better data to work with. "Using average tissue thickness is not very accurate because there's a lot of variation around 'average,' " explains Uberbacher. "If you put an 'average' nose on someone like me, for instance, I'd be unrecognizable," he says. The tissue-thickness data typically used for facial reconstruction come from a limited number of measurements taken from cadavers a half-century ago. By taking measurements from hundreds of recent magnetic resonance imaging scans of living volunteers, the ORNL researchers began to develop detailed predictions that consider age, gender, and ethnic origins.
Next, instead of basing the reconstruction on only a dozen or so points, they used the computer to plot thousands of points so the surfaces of the face would be mathematically based on the shape of the entire skull, not just a few landmark points directly below the skin.
Even the computer became more accurate over time, learning from its experience how to better predict a face. "We can program the computer to combine information to predict an outcome in much the way a child learns," Uberbacher explains. "It's like learning the difference between a cat and a dog. A one-year-old child often gets the two confused. That's understandable—four legs, tail, fur, ears. With experience, though, the child gathers more information—shape, meow—and draws better conclusions." Using case-based reasoning, ORNL's computers can do the same thing, combining information from the database into a composite—matching this forehead and that jaw, say—to reach more accurate conclusions about the shape of the face of an unknown victim.
ORNL and the University
of Tennessee are working together to apply
advanced computing and artificial intelligence to rapidly
and inexpensively reconstruct faces of murder victims
from their skeletal remains. Starting with an
unidentified skull, the new technology is expected to predict the
most probable face for the victim; identify gender, race,
and age; match police photo books; and factor in
genetic markers or DNA sequences that could be used to
predict the victim's appearance.
Today, it takes supercomputers like ORNL's Intel Paragon XP/S 150 to sift through gigabytes of data and create a three-dimensional image. But as the system is refined, Uberbacher, Mann, Mural, and Marks expect the time and cost of computerized reconstruction to drop dramatically: from a couple of months and $2500 to a couple of hours and $25. That would allow facial reconstruction to be widely used on unsolved crimes. New virtual reality features that allow forensics experts to add or subtract 20 years or 20 pounds with a keystroke should soon make results even more realistic. And remote access, by which scans of a skull can be e-mailed for face reconstruction, will make the technique available even to smaller police departments.
| The dark and sinister business of crime will never be shut down. But by developing techniques and equipment to identify victims, protect police officers, and bear witness against criminals, ORNL is helping to make that business riskier and less attractive to criminals. |
The dark and sinister business of crime will never be shut down. But by developing techniques and equipment to identify victims, protect police officers, and bear witness against criminals, ORNL is helping to make that business riskier and less attractive to criminals.
As Holmes would say to Watson, the game is afoot; it always will be. But the odds for the good guys are getting better.
Pete Xiques, a division manager at the Oak Ridge office of the Science Applications International Corporation, is a freelance writer in his spare time. He can be reached at peter.j.xiques@cpmx.saic.com
[ Next article | Search | Mail | Contents | Review Home Page | ORNL Home Page ]