Draw and order

The practice of using police composite drawings has not changed much in the 125 years it has been used to help identify perpetrators of serious crimes. Though practitioners of forensic composite art are skilled, the craft is still inexact at best. 

That’s not always a bad thing. Forensic art is not about creating something perfect or beautiful. Rather, it’s about developing evidence that was previously intangible. 

“Forensic artists should never be judged by how close they got to the actual person-of-interest,” says Suzanne Lowe Birdwell, a forensic artist with the Texas Rangers Evidential Art and Facial Identification Unit in Austin. Instead, she says, the goal is for the composite drawings to induce recognition of the subjects in the drawings, not to create museum-quality portraits of the suspect.

“The main purpose of forensic art is to generate leads for investigators, and this is possible even if every part of the drawing does not end up 100 percent accurate,” Birdwell says. “Remember, the artists don’t know what they’re drawing.”

Although composite drawings have been in use since the late 1800s, there are still relatively few full-time positions available for forensic artists in domestic law enforcement agencies. There are approximately 35 to 40 forensic art units across the country, employing between 60 to 70 full-time forensic artists in local, state, and federal agencies, according to Birdwell. Some forensic artists provide services on a freelance basis, on a volunteer basis, or as part of their duties in a crime analyst position if they have artistic skill and have been trained in composite drawing.

Non-profit service groups also employ forensic artists. The National Center for Missing & Exploited Children, for instance, has a staff of four full-time forensic artists who age-progress available early photos of long-term missing kids. Occasionally, the Center’s artists also attempt to reconstruct facial features from skulls of deceased, unidentified children.

About face

Regular forensic sketches are usually developed in a couple of hours, while the forensic artist interviews the victim or witness about what is likely the worst event they have ever experienced. For Birdwell, drawing for law enforcement is structured differently than normal “real” art, in that creativity is not used when making a sketch of a suspect. Forensic artists must suppress their creativity and develop the image based strictly on the results of the interview with the witness or the victim, according to descriptions and adjustments provided.

“The artists are also at the mercy of a myriad of situational factors that may negatively affect witness recollections, such as the timing, angle, and lighting at the crime scene,” she adds.

On top of that, the whole forensic art process is only successful if the composite sketch is published and is eventually seen by someone who believes they recognize the person in the sketch, and who then decides to come forward and volunteer information.

Nevertheless, Birdwell believes the value of forensic artistic images continues to grow, provided they are created by skilled, well-trained professionals, and developed in conjunction with a thorough interview of a witness/victim who maintains a reasonable memory of the event.

Let’s face it

Birdwell advises that the key to attaining a good likeness in the form of a composite drawing is to engage the witness or victim in a thorough interview.

“The composite drawing interviewing process is not like regular police interrogations,” she says. “[These] take a softer, lengthier approach, called a cognitive interview, which implores the witness to relax, enabling memory to flow more freely.”

Birdwell notes the best forensic art interviews encourage the use of the senses as the witness recalls details of the event in question, and the description of the suspect.
“The memories of witnesses are constantly being encoded through all of their senses, and although the artist’s goal is to develop a visual representation of the suspect via the victim’s memory, those memories may not have all been visual,” she says.

For example, the victim may know the suspect had stubble on his face because she felt it but never actually saw it, which is a tactile memory. Or she may have smelled alcohol or cigarettes (olfactory memory) or heard the muffler of a car (auditory memory).

“Having the victim recount all of these memories greatly increases the thoroughness of the description of the entire event, resulting in a more accurate likeness of the suspect’s face,” adds Birdwell.

The interview process is best conducted in a private setting where the forensic artist can be alone with the victim and take as much time as necessary to complete and obtain the drawing.

Face facts

When some police agencies don’t have access to or can’t afford a forensic artist, they sometimes opt for a commercial computerized sketching system. These work up to a point, but some programs only hold a limited number of facial features from which the witness must choose when constructing the suspect’s face.

“By contrast, a freehand forensic composite artist can literally add anything to the image, and can also make even the smallest of changes the witness may like to include,” Birdwell says. “A competent forensic artist never wants to limit her witness’ statements by not being able to fully develop the image the witness holds in her mind.”

Another downside to some commercial systems, she continues, is that many of the features used in these systems are photo-realistic. A composite drawing of a person-of-interest is best if it still retains some sketch-like qualities, because that broadens the interpretation of that face by the viewer.

The more “photographic” the image looks, the more the public views it to be ‘real’. The problem with images being too realistic is that viewers may eliminate possible persons in their minds as being a match to the drawing because every detail they hold in memory may not be exactly like the too-real image.

On the other hand, if the composite drawing is viewed as a sketch, the public understands it is merely a suggestion of the suspect’s appearance. “Humans recognize faces holistically,” Birdwell says. Which is to say humans take in the image as a whole, without measuring small distances between features.

Sometimes police agencies that purchase a software sketching package may not have a person who has been trained on both the drawing process or cognitive interviewing because the agency seldom needs composite drawings. Instead, the agency calls someone to run the software when they need a drawing.

One problem with this approach, according to Birdwell, is that the victim may not get the benefit of the detailed interview necessary to probe all areas of memory, including tactile, auditory, and olfactory, which may result in a less detailed rendering of the face.
The solution of course would be to open up more full-time forensic art positions—an unlikely prospect in an era of diminishing budgets, no matter how valuable the services might be.

“These services need not be limited to composite drawings of persons-of-interest, but should also include other genres of forensic art, such as age-progressed and age-regressed facial images, courtroom exhibits, image clarifications from video, and postmortem artwork that reconstructs facial features from skeletal remains to help identify unknown decedents,” says Birdwell.

A new wrinkle

One new wrinkle has been added to the forensic art process that promises to give composite drawings even more forensic utility.

A team of researchers at Michigan State University has developed software that takes facial composites drawn by forensic artists, and then automatically searches photographic mug shots stored in law enforcement databases for matches, significantly accelerating the suspect identification process.

“The forensic composite procedure used [publishing a sketch and waiting for volunteers to come forward] to identify a suspect is slow and tedious, and there is no guarantee of identifying the suspect,” says Anil Jain, a Michigan State professor of Computer Science and Engineering.

He says it would be desirable if police could automatically compare the suspect’s composite facial sketch drawn by the forensic artist to the millions of mug shots in state and federal law enforcement databases. The problem is current facial recognition software systems do not perform well in matching a sketch to a photo.

Jain believes the Michigan State research, which is supported by the National Institute of Justice, is the first large-scale experiment that attempts to match forensic sketches to police photographs. The system was originally developed by a former Jain graduate student, Brendan Klare, now at Noblis. Current enhancements are being carried out by graduate student Scott Klum.

“To our knowledge, there is no sketch-to-photo system available on the market,” says Jain, adding that all commercial face recognition systems work well when matching photos-to-photos, but fail miserably in matching sketches-to-photos.

The novelty of the Michigan State approach is to find image features that are amenable to matching sketch-to-photo. In order to achieve this, the Michigan State team first trained the system using sets of sketches and photo mug shots obtained from police from previously solved cases that involved both a sketch and a mug shot.

“Our system learned the similarities between a sketch and a photo based on these solved cases,” comments Jain.

Results have been promising. Using a database of more than 10,000 mug shot photos, 45 percent of the time the system matched the correct person.

In fact, according to Jain, the system performs even better than state-of-the-art photo-to-photo matching systems in finding hits.

He now has a prototype of the system, called the FaceSketch ID System, which is currently being evaluated in the field by the Pinellas County, Mich., Sheriff’s Office and the Michigan State Police—the two agencies that gave the Michigan State researchers access to sketch-mug shot photo pairs of solved cases used to train the FaceSketch ID system.

And Birdwell believes Michigan State researchers may be on to something useful.

“It may become one more tool by which law enforcement can attempt to identify a person-of-interest,” she says. The system, she notes, doesn’t appear to be endangering any part of the investigative process by doing so, which is a positive step.