Hidden gems

     If only life would imitate art — or at least a big-budget movie. Because if it did, pulling information off surveillance and security video would be a whole lot easier. Instead, law enforcement personnel routinely contend with bad camera angles; poor-quality video that is too dark, too shaky or runs too fast; obscured subject matter; lens distortion; glare; noise and other image-compromising issues.

     Then there's the problem of digital video evidence, says Paul Hartzell, digital evidence specialist for the Hennepin County Attorney's Office, located in Minneapolis, Minnesota. "The system recording the video is most likely proprietary, meaning users need specific player software in order to view it," he explains.

     Doug Perkins, director of sales for Ocean Systems, a Burtonsville, Maryland, manufacturer of video forensic systems, says that at last count there were more than 900 different digital proprietary file formats and zero universal players that allow folks to view the video, regardless of the surveillance system that created it.

     The ability to retrieve accurate information from surveillance and security video is essential to understanding the crime scene, apprehending suspects, and seeing that justice is served. Fortunately, when viewing problems do occur, law enforcement agencies can get a big assist from a variety of video forensic tools.

     Countless examples of the utility of video forensics tools can be found in law enforcement. Robert Sanderson, president of Audio Video Labs, a New York-based company that provides video forensic services to law enforcement agencies, lawyers, insurance companies and private parties, recalls a case involving a gang shooting in a crowded warehouse-turned-nightclub.

     "The camera was concealed by some type of wire mesh, [which] made it difficult to see the shooting," he explains. "However, once enhanced, the video rendered detail that was (previously) unseen and showed more than one shooter."

     Or, take the rape case where Sanderson was hired by the defendant.

     "Since this allegedly occurred in the rear of a convenience store, the cameras, which were facing toward the front, didn't catch the action," he says. "I enhanced the reflection in the front window and also revealed the plaintiff handing the defendant her phone number at the edge of the screen after the event."

     A video forensic system developed by Cognitech Inc. was used in the Reginald Denny beating case, says Lenny Rudin, CEO and founder of the Pasadena, California, company. Their hardware and software enabled investigators to enhance a video footage shot from a helicopter to identify a rose tattoo that, in the original footage, was very blurry and quite tiny, thus leading to the successful prosecution of the accused.

     Perhaps one of the most famous cases solved by video forensics was that of the Olympic Park bomber. Intergraph Corporation, based in Huntsville, Alabama, worked for months in conjunction with NASA to develop technology that would later identify the perpetrator of that crime, says Gene Grindstaff, executive manager and chief scientist of video tools. He vividly recalls the surveillance footage they had to work with.

     "All the lights had been cut off. But an NBC camera was still recording when the bomb went off," he recalls. "It was very dark, very grainy footage, and people were knocking into the camera. We took the analogue tape, and exactly registered multiple frames — there were about 120 frames. We could pull out the backpack and the logo on it, and we were able to see that it was one commonly used by mountaineers or hikers in North Carolina.

     "We were able to track this backpack to the region where it was sold," he continues. "We were also able to pull out a shoe, and see that it was a Size 9. All this and other evidence resulted in the suspect's eventual capture."

     It's easy to forget how often our activities are recorded — Grindstaff says the average person in the United States is caught on video approximately 30 times a day.

     "In order to properly investigate the majority of crimes, it is now necessary to be able to examine and enhance video from a variety of formats and sources," says Grindstaff. "This technology allows investigators to preserve, evaluate and enhance video evidence. Without it, evidence might be overlooked or lost."

     Improper retrieval of video evidence can limit image detail by as much as 20 percent, says Sanderson. "Repeated playing of VHS tapes, especially old ones that have been recorded-over many times, will reduce detail and create noise and dropouts with each play," he explains. "There is a tipping point beyond which the true value of the evidence is lost."

     To answer the need for video forensics, smaller agencies lacking someone trained in video forensics or the budget to create their own video forensic labs had to outsource forensic video either to a lab or to the next highest authority with video forensic capability, says Laura Teodosio, founder of Salient Stills Inc., a video forensics and image enhancement software company located in Boston, Massachusetts. In some cases departments lost control of the video and experienced processing delays. This is why even smaller agencies are starting to explore ways to process video in-house.

     The Rhode Island State Police formed a video forensics lab several years ago. Its responsibilities include processing the agency's own video as well as performing work for other departments, says Det. Lt. Dennis Pincince, who is in charge of the department's criminal investigations unit. In the beginning, department officials fielded approximately two to five requests for assistance from other agencies each day, but as more departments have added their own systems, the number of requests has dropped to two to five a week.

     The Corpus Christi (Texas) Police Department brought this function in-house a little under a year ago, says Michael Alanis, senior officer in the auto theft division. This move has resulted in greater efficiencies. "We don't have trouble viewing the video from different sources," he says. "We can make still prints and provide media outlets with more usable video to help locate suspects.

     "Prior to this, the media often wasn't able to play the video because they lacked the system to run it," Alanis continues. "Now, we can convert it to a Windows media file and e-mail it to them, and they can run on the news that night. We also can receive a video and have it ready for detectives to review the same day."

     When creating a video forensics lab, agencies must define their needs and set up the lab accordingly, says Hartzell. For example, if a department is outsourcing, what kinds of projects are being outsourced? He poses other questions to examine, such as:

• Does the department need it to build a case, or just to get something in the hands of cops looking for a suspect?
• Does the agency want to be able to analyze the video, or just make it viewable?
• Are officials seeing primarily VHS video or are they getting more DVR?
• Does the department want to be able to edit or produce video for other purposes; such as training, or are plans to simply manage surveillance video?
• Will everyone in the agency handle this task or just a few people?
• Is there a requirement that the lab be ASCLD-accredited? "If so, there will be guidelines on setting up a lab beyond the basic setup," he explains.
• Are there shared resources that can assist with analysis or processing?
• Does the department have access to grant or forfeiture money? "If money is not a problem, having more than one system would be beneficial," says Hartzell. "It's like a toolbox; one system may have a better tool for a specific process than another. Or at the very least, results can be compared between systems."

     And of course, departments must consider core equipment needs, Hartzell says. This includes setting up a dedicated PC that allows frequent software installations. A Macintosh, he says, may or may not work with all software; but anything not specifically written for a Mac should open on a PC.

     "The setup can be as simple as using the PC's print screen function to capture an image to the OS clipboard, or as complicated as having a complete setup that includes a video system with a capture card," he adds. "There are many products available that offer a complete video forensic solution, ranging in price from several hundred to several thousand dollars."

     Agencies must also budget money for training. For example, though most video forensic systems providers offer no-cost training, agencies still must budget to send personnel to a specific location to receive it. This is only part of the puzzle, Hartzell adds, noting those working in video forensics also need training in video evidence analysis as well as evidentiary concerns.

     There are additional training avenues. For example, the Law Enforcement & Emergency Services Video Association (LEVA), found at www.leva.org, can help departments learn to handle processing in-house and aid their understanding of admissibility issues, Hartzell says. Other resources he mentions include Wikipedia (http://en.wikipedia.org/wiki/Forensic_video) and online forums, Web sites, etc. that are dedicated to sharing this kind of knowledge.

     The important thing is to thoroughly research the systems and their capabilities, Hartzell says. But even here, training helps.

     "Many manufacturers of video forensic equipment will allow potential customers to test their products," he says. "The downside is that most require training before officials can fully appreciate what [these systems] can do."

     Pamela Mills-Senn is a freelance writer based in Long Beach, California. She specializes in writing about public safety issues. She can be reached at [email protected].

     There are a number of video solutions available. The following lists some of the more common systems in use by law enforcement today.

     Cognitech Inc. offers court-approved forensic video/image processing and forensic video measure software with its Tri-Suite "VIVA-AM," which consists of:

     Cognitech Video Investigator (hardware and software): The system offers hundreds of plug-ins from unique image restoration/super resolution tools that take users from a point where it is impossible to discern features or details due to blur, noise or low resolution, to where a face or license plate can be positively identified.

     Video Investigator (hardware and software): This system was first used to identify four assailants in the Reginald Denny riot case. The tool enabled investigators to identify a rose tattoo that was integral to the case.

     Video Active (software): This software allows users to simply pop a tape or DVR recording into a computer and have the footage show up demultiplexed, enhanced and enlarged, with the noise and blur removed. This is all done in real time; users see the demultiplexed and enhanced images at the same speed as the tape plays. Video is captured 100 percent uncompressed, but storage takes only 50 percent or less due to a proprietary "video-zip" process.

     Auto Measure (software): This forensic software allows identification of individuals and objects not only based on their image, but on the dimensions of that image, such as the width of a shoulder, someone's height, the length of a car, etc. It is fully automatic, including correction of fish-eye lens distortions. The user just clicks on the points he wants to measure, and the measurements are made automatically.

     Intergraph Corporation offers a set of video forensic software tools through its Video Analyst product. This software, which sits on top of Adobe Premier Pro, allows users to capture, enhance and analyze surveillance video, says Gene Grindstaff (currently, the company has exclusive access to NASA's VISAR technology, which came out of the Hubble telescope and other weather satellite enhancement technologies, and is used to enhance multiple frames). Key features of the system include the ability to:

• Automatically register multiple frames and average them together, resulting in a super-resolution image that reveals details that would otherwise be missed.
• Demultiplex video without dropping a frame, accomplished via patented technology unique to their system, says Grindstaff. It can handle up to 128 multiplex cameras. (A real-time version is also available to allow the viewing, demultiplexing and recording of a multiplex tape using a standard VCR.)
• To remove blur, smear, noise and enhance either single or multiple frames.

     "All of these features are fully integrated into Adobe Premier Pro," says Grindstaff. "To use the features, the user merely captures the video segment or imports it from a digital file and places it on the timeline. They just drag the desired tool from the menu and drop it on the video. After a few adjustments, the video is enhanced."

     Ocean Systems. One video forensic solution from this company is dTective, a suite of tools that can run both inside and outside the Avid system — depending on the tools, says Doug Perkins. Included in this system (but not limited to) are:

• dPlex Pro: Deplexes videos taken from multiple cameras. Users get each camera view broken out, allowing them to play images from just one camera, or multiple cameras simultaneously.
• dPlex: Allows users to find one nuance that is different in very similar images, and pull images from that camera only, ensuring that they are not mixing in images from other cameras.
• dVeloper: A frame averaging process that enhances images, eliminates background noise and so on, allowing viewers to discern things like a license plate, tattoos or jewelry, etc.
• arithMATIC: Users can perform any sort of arithmetic function, such as adding, subtracting or multiplying. Let's say, someone has stolen six cartons of cigarettes from behind a counter, only investigators don't know this. The software compares before and after shots, determines that the cartons are missing and highlights the differences between the before and after images.
• DVR dCoder: Designed to solve the problem of working with proprietary video. It will capture images as uncompressed TIFF or AVI file formats and allow users to decide where they want to go to read it (Avid? Photoshop? Quicktime?).

     Ocean Systems also has Clear ID 2.0, which runs off Adobe Photoshop. This is designed to perform digital image enhancement (it emulates LEVA's best practices for digital image enhancement). Clear ID will do things like pattern removal, deblur images, remove artifacts, and so on. This is not included in the dTective suite, but must be purchased separately.

     Salient Stills. VideoFOCUS Pro v3.0 (VFPro 3.0) is an upgrade to the company's flagship video forensic software, VideoFOCUS Pro. The new features are designed to enable users to process analogue and digital video from start to finish, within a single system, says Laura Teodosio, CEO. Some features include:

• Analogue video capture: The software stores multiple hours of video for review, image creation and sharing and archiving. Once digitized, tapes can be safely stored.
• Digital video importing and transcoding: This feature allows users to easily work with a range of file types, including AVI, MOV, WMF and ASF, and streaming video formats, as well as cell phone movies.
• Digital video screen capture: Users can capture video data directly from proprietary DVR movies then play on screen.
• Quick scan and search: This feature allows users to quickly review and mark areas of interest in a long video sequence.

     The software also offers demultiplexing capabilities; drag and zoom for stills and video, for better inspection of small areas; new filters for clearer imagery; extended filtering of single images and entire videos, including tracked areas, so users can easily perform operations like histogram equalization, levels and color space adjustments, sharpen blur and video stabilization; and editing tools that allow users to cut, copy and paste, seamlessly composing new videos. Additionally, VFPro 3.0 maintains a complete record of the operations performed, creating an audit trail.

     VFPro 3.0 runs off a standard PC or laptop and doesn't require proprietary hardware, says Teodosio.