Workshop: Digital Imaging Seminar for Law Enforcement
Hosted by: FBI Forensic Science Training Unit
Place: Golden Nuggett Hotel Las Vegas, NV
Time: May 12-16, 1997
Title: Forensic Video Image Analysis
Author: Thomas R. Edwards
Affiliation: TREC, Inc
Holmes Avenue NW
E-mail address: firstname.lastname@example.org
web site: http://www.trec.com
Forensic Video Image Analysis is a new scientific tool for perpetrator enhancement and identification in poorly recorded crime scene situations. Forensic Video Image Analysis is emerging technology for law enforcement, industrial security, and surveillance. Forensic Video Image Analysis addresses the following problems often found in these poor quality video recorded incidences.
Quad, nine, camera scenes per single video frame
Perpetrator too small to identify or too distant from the camera
Motion blur due to perpetrator running through the field of view
Image blur from lens poorly focused or coated with dirt, dust and grime
Poor alphanumeric legibility, e.g., license plate, logos, patches, etc.
Corrupted video fields/frames from seldom adjusted/cleaned VCRs
Field recorded video played back on frame display equipment
Ghost perpetrators within frames, i.e. recorded in one field only
Wavy distortion, video tear, poor sync signals, time base distortion
Lack of intensity and contrast due to over used, poor quality video tape
Dark noisy environments, low light levels, poorly lit perpetrators,black on black
Incorrect or no azimuth on playback VCRs for azimuth recorded VCR tapes
Multiplexed Time Stamped/Camera Number decoding errors, 4 to 16+ cameras
Blocky/blurry enlargement of the areas of interest in the computer; X2, X4, X8
The Acceptance of Results
These problems have been solved and the results demonstrated in a number of actual video recorded crime scene cases (references available upon request where appropriate). Investigators have obtained positive identifications and arrests. Convictions have been obtained using this new technology. Public Defenders have made practical use of this technology for their clients. Forensic Video Image Analysis, as developed by TREC, has been accepted by law enforcement and the courts in various jurisdictions through out the country.
Forensic Video Image Analysis can be defined as the relationship between law enforcement, industrial security, surveillance, digital image processing and the courts.
When an investigator first contacts TREC, a report and all supporting incident documents are requested, along with the original crime scene video tape. Copy tapes are declined due to loss of image quality, unless there exists no original. The report describes the nature of the incident and the areas of interest or time periods that require Forensic Video Image Analysis.
The first step in the analysis process is to play the crime scene VCR tape in a suite of VCRs. This procedure discovers which playback VCR provides the best signal to noise ratio, i.e., the best image. If the videotape is multiplexed encoded, the signal is routed through the appropriate decoding unit (there appears only three distinctly different multiplex formats in the industry).
The suite of VCRs has jog/shuttle, frame and field tape positioning control, time lapse between frames and azimuth selection, tracking, VHS and SVHS inputs and outputs.
The route of the video signal in all the above is from the VCR through a full frame digital time base correction (TBC) unit with color adjustments plus black and white brightness and contrast control (256 steps), freeze frame, freeze three selectable fields, and a number of other supporting features. Either through cabling connections or via the TBC, the color burst signal is eliminated from the composite video signal. Color images are lower resolution than black and white images. Forensic Video Image Analysis requires the highest resolution images for the greatest amount of detail. The TBC is an essential aspect of the signal-conditioning path. As a complement to the hardware TBC, software time base correction allows further adjustment of misaligned video.
Once the best playback VCR and appropriate demultiplexer unit have been discovered, each selected frame of video is converted into a digital format of 480 lines, 512 points per line and 256 shades of gray per point or pixel (picture element). Images are stored in TIF format, tagged image file format.
TRECWare, Forensic Video Image Analysis software is applied to the selected frames that have been captured in the computer. Each image requires a unique sequence of processing steps for optimal results. Resolution enhancement, deblurring or sharpening, along with intensity adjustments are applied and where appropriate noise filtering. Nighttime scenes are turned into daylight scenes. Black on black images become clearer. A number of non-linear intensity transformations, from logarithmic, exponential, to stretching with clipping have proven very worthwhile. One of the most significant features of TRECWare is the optical quality enlargement, magnification times two, of any quarter section of the captured image, to a full frame size. This feature magnifies a quad camera section to a full frame image. TRECWare is capable of splitting a frame into the even and odd fields of 240 video lines. Either field can then be vertically enlarged into a full frame (480 video lines) with optical quality characteristics. This latter feature is a solution to the classical problem of perpetrator motion blur, i.e., perpetrator runs fast enough in front of the camera to cause image blur preventing feature identification. This feature is also a solution to image jitter. Complementing the overall and vertical magnifications times two is horizontal magnification times two. This latter enlargement allows more white space between characters for enhanced legibility, i.e., license plate alphanumerics. Successful results have been obtained with TRECWares digital magnification up to times eight.
Hardcopy results are obtained via a high resolution laser jet printer, 2400 dots per inch, 256 shades of gray, near photographic quality, a printer screen frequency of 212 line pairs per inch (half tone). These prints are inexpensive, approximately three cents per copy. Photographic quality hardcopy is obtained from 8.5" by 11" dye sublimation hardcopy, a printer screen frequency of 300 line pairs per inch (continuous tone). Dye sublimation prints are excellent for court room presentations, but at $2.50 per copy expensive. The highest resolution output is 35mm slides. For large screen presentation of images in the courtroom slides are the best. Overhead film projection of images is less acceptable and significantly more expensive. Both VHS and SVHS videotapes can be produced. One aspect of the video tape production is the slowdown capability of fast moving sequences for easier comprehension of the original scenes or selection of just the camera of interest in the multiplexed VCR tapes. The processed images can also be prepared on videotape, each image displayed by a user selectable number of seconds. On 32" large screen monitors with a field advance VCR played through a full frame TBC unit, court room presentations are excellent. Two or three 32" monitors played through a very wide bandpass distribution amplifier satisfy the court room requirements for jury, judge, and opposing counsel viewing. The images are stable, very bright, and displayable for long periods of time.
Access to the Technology
Forensic Video Image Analysis is available in two fashions, VIP Service and SPvI Workstations. VIP Service is 24 hour turn around of Video Image Processing Analysis via overnight airborne express delivery systems. Via fax, TREC provides a set of instructions for investigators regarding the video-recorded tapes that require Forensic Video Image Analysis. A SPvI Workstation is a complete Software Processing of video Information laboratory environment for those agencies with technical services officers dedicated to audio and video analysis. A SPvI Workstation is installed with four days of on site training. There are three additional courses of instruction for the analysts as skills in this field advance. The technical service officer is encouraged to be a member of NATIA, National Technical Investigators Association. A SPvI Workstation consists of multiple VCRs, TBCs, High Resolution Monitors, Dye Sublimation and Enhanced Laser Jet Printers, High Resolution Scanners, and a sophisticated microprocessor system for data gathering and image analysis. The heart of Forensic Video Image Analysis is coded in a software program known as TRECWare.
Forensic Video Image Analysis allows cost savings to the criminal justice system by plea bargains in lieu of jury trials. No one recognizes themselves better, even in very poor quality though Forensic Video Image Enhanced results, than the perpetrator. Such recognition leads to plea bargains, a significantly less costly undertaking than a full trial with jury.
A number of Federal, State, and local agencies have installed SPvI Workstations. Numerous agencies, departments, offices, and industrial security organizations throughout the country have utilized VIP Service.
Copyright © 1998-2010 TREC All Rights Reserved Last Updated Sunday, August 29, 2010