The Boston Marathon bombings on 15 April 2013 put a spotlight on photonics-based technology that could play a central role in improving efficiency of such an investigation, and possibly help prevent similar incidents.
For example, matching face imagery acquired from video or still cameras with face-recognition databases is crucial. Today, suspects can be identified just by their gait caught on camera. A suspect's pulse can be calculated from video, suggesting their shifting emotional state as they anticipate an action. Explosives can be identified from as much as 100 meters away.
Counterterrorism officials would try to match faces of possible suspects with faces in databases of images from passports, visas and driver's licenses, using facial recognition software. Hundreds of photos and videos were under study in Boston, frame by frame, as agencies investigated the bombings, which left at least three people dead and as many as 170 people injured. Officials asked the public to offer more video or photos of the area.
At BAE Systems, in Burlington, MA, about 15 miles northwest of the actual marathon route, Dr. Stephen DelMarco, Senior Principal Research Engineer, knew that facial recognition technology would be central to the forensic investigation.
At SPIE Defense, Security and Sensing 2013 in Baltimore, DelMarco presented developments in fusion techniques, still in the research stage, that will make storing and transmitting facial images much more efficient. His system employs multichannel image fusion for 3-D color face imagery, offering a big step up from previous matching systems.
More channels, more data
"Now you can exploit multiple channels -- for example, red, green and blue color channels, along with depth information that give you better image recognition," he said. "In the old days, you had a single channel; a gray-scale image to process, which provided less information."
One benefit will be image compression for transmission of images captured by color cameras over wireless devices or for storage on disk drives. The reduced size of compressed images offers the same image quality but in a much smaller space, a more economical approach that also, by making the image smaller, allows transmission over limited-bandwidth channels and reduces transmission costs. "You can make something happen that you couldn't do before," Dr. DelMarco said.
"What we've seen in Europe, in London, with the ubiquity of cellphones and security cameras, is that everything is on video or still imagery. There's a massive amount of information -- but the big thing is the accuracy of it. If you are trying to recognize somebody, you want it to work well, with limited false alarms and missed detections. You want to find the right person." The keys to accuracy, DelMarco said, include improved resolution, advances in algorithms for effective matching and increased computational power.
In Texas, Dr. Sos Agaian heard the news from Boston at the University of Texas at San Antonio. His first reaction was that investigators could benefit from integrating and enhancing information from many kinds of mobile devices.
"Visually, when you enhance the image you can see things that at first you didn't see," said Dr. Agaian, who is the Peter T. Flawn Professor of Electrical and Computer Engineering at UTSA.
If you have several cellphone images, you can integrate them all into a huge panorama of the scene of an event. And you can make faces and actions more recognizable.
In Agaian's lab, he can calculate someone's pulse by recording their face with a camera. A few minutes later he will again calculate their pulse.
"At first they may appear to be very calm," he said, "showing no emotions. Then, the pulse can show something happening to them that you didn't see by their actions. You can ‘see' what the person is feeling, as their heart starts to beat fast. It suggests something is going on that's having an emotional effect. You can analyze the emotion that they are trying not to show."
He said that 20 seconds of video can allow an accurate pulse reading. "Then if 10 minutes later I have another video, I can calculate another pulse and sense that there is something they are preparing to do."
The human voice plays a major role alongside still images and video, he said. After the Boston bombings, many witnesses will have captured such diverse information on cellphones, PDAs or iPhones. Agaian says his lab can detect precisely what cellphone or camera was used, even working from copies of the original.
Digital evidence recovered from a cellphone, he said, can provide information not only about the users but also about the accuracy of data -- still images, video, or voice -- and about the "DNA" of the mobile devices.
Dr. Agaian chaired the Mobile Multimedia/Image Processing, Security and Applications conference at the Baltimore SPIE meeting.
When Christopher C. Carter, program manager at the Applied Physics Lab at Johns Hopkins University, heard the news of the bombing, he knew immediately that forensic teams would be making use of optical techniques to check samples for explosives and other chemicals, whether at long distances or close up. An example is Raman scattering which provides a "fingerprint" of the molecules to be identified.
"Raman spectroscopy is being used by both military and civilian agencies to identify bulk chemicals," Carter said. "You'd use a laser to look at the Raman scattered light, from a few centimeters away, and detect precisely what is there. Other active optical techniques can identify explosives from up to 100 meters away."
Carter chaired a Spectroscopic Chemical Detection session, part of the Chemical, Biological, Radiological, Nuclear, and Explosives (CBRNE) Sensing XIV conference.
CBRNE sessions at DSS covered a range of chemical and explosive detection methods including active and passive hyperspectral imaging -- active implies a source, such as a laser, to stimulate a response from the chemical of interest, while passive techniques use photons from the background environment. Other applications include quantum cascade frequency-modulation (FM) spectroscopy, photoacoustic spectroscopy, laser-induced breakdown spectroscopy (LIBS), and surface-enhanced Raman spectroscopy (SERS).
Meanwhile, in Washington, a cautionary note came from Dr. Augustus Way Fountain, an expert in electro-optics relating to chemical, biological, radiological, nuclear and explosives (CBRNE) sensing. He said that the search continues for ways to detect explosive devices in advance of an event, and for now the investigation into the Boston Marathon bombing would rely on "traditional optical spectroscopic and forensic methods."
Fountain, who is chair of the CBRNE Conference at DSS, is the Senior Research Scientist for Chemistry in the Research and Technology Directorate, U.S. Army Edgewood Chemical Biological Center in Maryland. He also serves as an at-large U.S. representative to the NATO Sensors & Electronics Technology Panel, advising them on CBRNE detection.
"The Holy Grail of bulk or trace explosives detection remains elusive to both military and homeland defense agencies," he said in a statement. "There are a number of research and development efforts across agencies working to detect persons or vehicles containing an explosive device before it is detonated; many of these rely on photonics-based technologies. SPIE's DSS Conference in Baltimore is a unique opportunity for the people developing these technologies to present their latest developments in explosives detection."
Ford Burkhart is a Tucson-based freelance writer.