EVERY FACE TELLS A STORY
- in search of the missing and the suspect -
PART 2

By Karien v/d Merwe

• In October 1999 Supt De Lange asked Insp Jeanette Naudé to compile a facial reconstruction of a suspect in a theft case (Hercules CAS 34/9/99). At that time, Jeanette had “coincidentally” read an article in the Huisgenoot about a suspect in another country who was also allegedly committing acts of larceny on a grand scale. She noted that not only were the suspects’ modus operandi similar, they also had similar facial features. They were, however, not the same man. Nevertheless, Jeanette contacted Supt De Lange about the article and faxed the information to him. He forwarded the previously published article together with the information about the South African suspect to Huisgenoot, highlighting the remarkable similarities between the two cases. The follow-up article was published and two weeks later, Supt De Lange had his suspect.
   
• Before the Facial Identification Unit decentralized to other LCRCs in the province, the Pretoria Unit had to cover the entire former Transvaal. In this way they were often able to pinpoint links between crimes committed in various towns. Seasoned detectives will probably recall the term “odd couples” used to describe two-member team tricksters hitting one small town after another in quick succession. One such odd couple comprised a tall, gangly silver-tongued thief and his short, quiet, dumpling accomplice. They would enter small enterprises, especially pawn shops, and ask to exchange an amount of money for petty cash. Under the watchful gaze of the owner or manager, they would by sheer sleight of hand, con the unsuspecting victim out of a much larger amount in petty cash than the amount of money they gave in exchange. The two Pakistanis were eventually caught and deported.
   
• Although facial comparisons and composites form the main business of this Unit, they are sometimes challenged to extrapolate their expertise and resources to other objects. During the late 90s, a white man dressed in casuals nonchalantly walked into small branches of financial institutions, flipped open a black attaché case, pulled out a firearm and demanded money. He got away with it time and again. The astonished expressions on the faces of the security guard and cashier can clearly be seen on the security camera’s video footage. Facial reconstruction of the suspect was difficult because of the position of the camera. However, there was something inside the attaché case which lent itself to comparison: a book with a distinguishable cover. The detectives only took casual note of this “unlikely to be used” bit of detail and filed it in the docket. Some weeks later in a quirky twist of fate, this robber’s own children’s kindergarten teacher identified him after he had robbed a bank. The police were waiting for him at his home, where he was arrested. While searching the house, detectives spotted a familiar-looking book and an attaché case that also looked suspiciously similar to the one on the videotape. The detectives confiscated these two items as possible evidence and, with the help of the Identikit members, the investigators were able to state in court that a “similar book and similar attaché case” had been present on the security camera’s footage. What were the odds of that happening by chance? After the other members of the household had been eliminated as possible suspects, the detectives finally confronted the suspect who soon after submitted a formal admission of guilt. The Identikit members themselves did not give evidence in court, but served as a back-up service strengthening the circumstantial evidence in the case.

INFOBYTE: VISUAL MEMORY

In recent years there has been a radical reversal in scientific conceptions of how visual memory works. Previously, it was thought that the longer you stare at an image, the more you will remember about it. Both logically and intuitively, this conception would seem to make sense - it seemed logical that the greater amount of time an image has to impress itself in your memory, the better you should be able to recall it.

But this is one instance where common sense is wrong. It turns out that, as far as seeing goes, less is more, and a great deal less translates into a whole lot more. Research into visual intelligence has revealed an astonishing fact. As McCarthy writes in Mastering the Information Age, “The faster we go, the more information we can absorb.”

How can this be? Because of a simple “grouping” principle objects are easier to remember when seen in context than they are when seen alone. Like words and musical notes, which gain significance as part of a melody or story, individual objects take significance as part of a picture. For example, you are watching a movie on television. As the camera begins to pan across a scene, you can fill in the gaps and tell what the remainder of the shot is going to show. When you capture an image quickly with your eyes like a camera, your eyes are registering not just random objects, but objects in context. Looking at a picture, or a chart, or a page, each individual object is glimpsed within a larger context, enhancing comprehension and retention. Though the time spent looking at the image is brief, your picture smarts* immediately grasp the objects, relate them to each other, and store them in your visual memory banks.

While you consciously register only a few objects, subconsciously your visual intelligence is picking up far more. That’s why you can retain images better when you quickly open and close your eyes like a camera shutter than by staring at them trying to soak up every detail. “Power seeing” is a natural, effective technique for gaining visual knowledge and finding answers to visual problems quickly and effectively. It has helped many people to achieve success and is based on our picture smarts’ wonderful ability to mentally assemble a whole picture from incomplete parts. In fact, this portion of your hidden genius is designed to automatically find meaning and pattern in as few visual clues as possible and to fill in or transpose elements to reach closure.

Pattern recognition is another of those capacities hard-wired in your brain and inherited from the time when primitive man roamed the African and European grasslands. They had to be able to tell friend from enemy and predatory carnivores from harmless herd animals. From just a few dark splotches against a yellow background they had to put together the visual cues that spelled the difference between potential predator and potential food. - in a hurry, and from far away - before the intruder drew near enough to do damage or get away.

Your visual intelligence automatically -

• scans for and recognizes patterns;
• search those patterns for a larger meaning that could fit them within a picture; and
• seeks visual closure by generating a whole image from its available portions.

Here’s how it works. When new visual data - images or portions of images - come in through the senses, your visual intelligence immediately scours through millions of previously stored images, searching for matches. The raw sensory input about colour, shape, relationship of size and distance and so on is converted into images that are meaningful in the context of your experience. Your image smarts seem to enquire: “Is anything familiar here? Is there a pattern I can recognize?”

When you are asked to guess something like the next number in the sequence 565656565, you probably have no difficulty guessing “6". That’s because the mind is concerned with meanings. It doesn’t passively absorb one individual fact or image or word. Instead, it hunts for a pattern or context within which inputs have a larger significance. We remember a sports statistic, or a memorable scene in a movie, or a moving paragraph in a book, not because of the numbers themselves, or the individual image or the individual words, but because of the larger meaning that they convey to us within that context.

You can demonstrate for yourself just how this process works by reading the following paragraph. Don’t struggle to puzzle it out or read it word by word. Scan through it quickly, and you will discover the way your visual intelligence automatically fills in the missing letters and recognizes the whole word.

The phaomnnehil pweor of the hmuan mnid
Aorccdrnig to rscheearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the lteers in a wrod are, the olny iprmoetnt tihng is taht the frist and lsat ltteer be at the rghit pclae. The rset can be atotal mses and you can sitll raed it wouthit a prbelm. Tihs is bcuseae the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe. Amzanig huh?

If you are like most people, your image smarts shuffled the letters without much effort. Your visual intelligence searched for patterns and completed meanings. The result? Your understanding of the individual words and the paragraph as a whole was as clear as if every letter had been correct in the first place.

[*visual memory and visual intelligence]
Source: Stine. J. M. Super Brain Power. 2003. Judy Piatkus (Publishers) Limited. London.

Your visual intelligence automatically - scans for and recognizes patterns; search those patterns for a larger meaning that could fit them within a picture; and seeks visual closure by generating a whole image from its available portions. Here’s how it works. When new visual data - images or portions of images - come in through the senses, your visual intelligence immediately scours through millions of previously stored images, searching for matches. The raw sensory input about colour, shape, relationship of size and distance and so on is converted into images that are meaningful in the context of your experience.

MOST PROCESSES CAN BE COMPUTERIZED - SO WHAT ABOUT FACE RECOGNITION SOFTWARE?

Biometric intelligence
Biometric security should be seen as an extension of human intelligence, and not as a replacement for it, because automated security will only be as good as the human intelligence that backs it up. The danger of relying too heavily on technoloy is nowhere more real than in the area of biometric surveillance. Such surveillance is most effective if the people you are trying to locate are not aware of its use. Audit trails left by an individual as he or she uses airports, car rentals, and any other services that require biometric authentication (ie possibly any activity that requires the use of a credit card, driver’s licence, passport, or any other major form of identification) could become a significant contribution to intelligence systems.

New real time security alternatives are a reality today with the I-CUBE lightning-fast Face Recognition System. A leading developer of mission critical biometric solutions designed for easy integration into existing systems, I-CUBE is committed to creating new benchmarks for security. Their emphasis is on accuracy. They promise a full range of biometric authentication including -

• tracing of multiple faces simultaneously in real time;
• enrolment of facial images from a live or recorded video stream;
• enrolment of facial images from a static image (JPEG);
• identification (one-to-many authentication) from a video stream;
• verification (one-to-one authentication) from a video stream; and
• database search of static images for a specific individual.

The I-CUBE package includes the ability to link to a camera, a proximity card reader, a fingerprint reader, a relay output card, and demonstration applications. The system also provides a full OCX interface to allow you to integrate any component of the system into custom applications written in Microsoft Visual C++ or Visual Basic.

An interesting aspect of the system is that the I-CUBE FRS learns, remembers and recognizes, and adapts. HNET emulates the human brain in structure and function, becoming more familiar with your face each time it sees you, adjusting for difference as a result of ageing and cosmetics, without increasing the size of the biometric template.

Information provided by Barry T Dudley (MBA{IT}; MSc {Image Analysis}; Bsc {Brewing}; Bsc Hons {Waste Technology}
For more information about system requirements, features, speed, and accuracy, or a free evaluation version of the I-CUBE FRS, please contact Barry T Dudley at cell: 082 562 8225, or info@l-cube.co.za or www.i-cube.co.za
According to Barry, Crime Watch in KwaZulu-Natal is using the system with great success. We hope to bring SAPS Journal readers more information in a future issue.

Memory - not a static phenomenon

In her book, Portraits of Guilt, acclaimed American police artist Jeanne Boylan, quotes the findings of Elizabeth Loftus, a professor of psychology at the University of Washington, and discusses her application of the findings to interviewing a witness: “Elizabeth Loftus, a professor of psychology at the University of Washington, had done studies proving that memories are not static, that ideas can be implanted, recall skewed and images overridden if an eyewitness is exposed to suggestion. As I’d seen, in standard police practice witnesses are bombarded with mug shots and catalogues of facial photographs and asked to point out features that look something like what they remember. It’s like picking out your lost luggage by pointing to an airline baggage chart full of handles, fabrics, latches and locks. If only the human psyche were so simple. I took Loftus’s findings and used them in thousands of cases over many years, continually refining my interview methods to avoid leading the witness. I tried to make them feel safe, so that details could emerge from the unconscious. Meanwhile, I developed my artistic ability to translate a witness’s words into an image ... One could compare human memory to a glass of water with an object (eg a traumatic experience) submerged in it. One should allow the ripples to settle, and once the surface is less agitated, the embedded image becomes clearer. The way I work is to talk with a witness about pleasant or neutral topics. The only reference to the task at hand is the insertion every 15 or 20 minutes of a single quick question - always phrased in the present tense: ‘Should I draw a face that’s longer than it is wide, or should it be more round?’ Keeping the conversation based in the present helps to anchor the witness in the safety of the current moment. I never refer to the suspect or the crime. Instead, I only refer to shapes and colours and textures.”

Source:
Boylan, Jeanne. Portraits of Guilt. 2000. Pocket Books (A division of Simon & Schuster Inc). New York.

The SAPS Journal submitted the above to Insp Jeanette Naudé for her comment. She responded as follows:

“I agree only in part with the findings of Prof Loftus regarding interviewing a witness:
‘ ... memories are not static ... ideas can be implanted.’
This is true when the witness is not sure of the facial features. If the witness is sure about a feature, I do not believe you will be able to influence the witness. In the light of the Professor’s findings I find that most compilations are partly accurate, because some people’s ‘power seeing’ is better than that of others. It is important that the investigating officers use the facial composite with the foreknowledge about which parts of the face are accurate. For instance, the hairstyle and eyes could be accurate while the nose and mouth are not. The compiler can usually detect this and it is always important that he or she confers with the witness to make sure which parts are accurate. This information should then be given to the investigating officer.”

LCRC contact numbers where Identikit personnel are stationed:
Durban:(031) 310 3195
Johannesburg:(011) 198 8177
Bloemfontein:(051) 503 2853
George:(044) 803 4542
Krugersdorp:(011) 951 1802
Nelspruit:(013) 759 1191
Pietermaritzburg:(033) 845 8500
Polokwane:(015) 290 7158
Potchefstroom:(018) 299 7762
Pretoria:(012) 393 2035
Vredenburg: (022) 713 5222
Cape Town:(021) 467 6160
Witbank: (013) 655 5323

Acknowledgement:
The SAPS Journal would like to extend a special word of thanks to the members of the Facial Identification Unit, Pretoria, in particular Insp Jeanette Naudé who shared her knowledge, and to Dir John P Lambert, Provincial Commander, LCRCs Gauteng, who approved the article for publication.