Figure 1 Ghost Perpetrator
An Aspect Ratio Problem and Solution
for LaserJet Printers
by
Thomas R. Edwards and Alfred A. Saccone, Jr.
Introduction
Forensic Video Image Analysis, as an emerging forensic science in law
enforcement, industrial security, surveillance, and the courts needs to satisfy the
requirements of scientific evidence in the courtroom. In particular, the images of a video
recorded crime as seen in the courtroom, those on the television monitor, and the printed
hardcopy must all present the same view with no distortion. Forensic Video Image Analysis
must be held to the same stringent requirement as the Forensic Photographer, "a fair
and accurate representation of the crime scene."
There are currently two tests for the acceptance of Scientific Evidence in the Courtroom, the Frye and Daubert tests. Frye requires the general acceptance of the technology in the particular field in which it belongs. Daubert is a scientific reliability test which replaces the Frye test in Federal civil and criminal cases, and is slowly becoming the standard in State cases.
Both tests require that the enhanced images be a fair and accurate representation of the original images, the same basic requirement, which has been placed upon the Forensic Photographer.
This report discusses a vertical elongation distortion problem and its solution in LaserJet printed video images. The problem is the result of the 4 / 3 aspect ratio inherent in television signals.
The problem exists because video dye sublimation hardcopy, though photographic quality and lacking any distortion problem, is very expensive. LaserJet hardcopy on high-resolution special purpose paper is inexpensive and provides a near photographic quality image, though previously suffering vertical elongation distortion. Budgets dictate LaserJet hardcopy be utilized wherever possible.
Consider a real example of the problem. The perpetrator in the Figure 1 is in every other video line, caught in just one field of a video frame. The perpetrator appears as a Ghost in the video image.
Figure
1 Ghost Perpetrator
Figure
1 Ghost Perpetrator
The images were captured by an Imaging Technology Inc, frame grabber after the video signal passed from the VCR thorough a full frame time base correction unit. Only the luminance signal was used in the analysis, chrominance was 75 ohm terminated. The images were processed in the computer by TRECWare, TREC’s Forensic Video Image Analysis software.
Split the frame into the fields and view
both fields, the field that does contain
the perpetrator, Figure 2 and the field that does not
contain
the perpetrator, Figure 3.
Figure 2 Odd Video Lines - One
Field
Figure 3 Even Video Lines - One Field
Figure
2 Odd Video Lines - One
Field
Figure 3 Even Video Lines - One Field
The time stamp between the two fields shows that the images are three seconds apart. 00:52:25 is the time stamp on Figure 3 00:52:28 is the time stamp on Figure 2. How this particular image evolved in this manner is unknown. Only the persistent viewing of the video by the investigator allowed detection of what has become known as the "Ghost Perpetrator", Figure 4. The perpetrator is just half height, suffers time base distortion and vertical elongation, the topic of this report. Figure 3 shows the field that does not contain the perpetrator.
Figure
4 Ghost Perpetrator
Figure 4 Ghost Perpetrator
The Problem
The aspect ratio problem becomes apparent when the field containing the perpetrator is vertically magnified to a full frame, Figure 5. Note the vertical magnification is without aliasing. The field of 240 video lines is vertically enlarged into a full frame of 480 video lines with optical quality characteristics.
Figure 5 Vertical Magnification Field to Full Frame (240 to 480 video
lines)
;
Figure 5
Vertical Magnification Field to Full Frame (240 to 480 video
lines)
The perpetrator in Figure 5, a computer printer hardcopy TIFF format file appears to suffer a vertical elongation. The perpetrator appears to be stretched vertically making him appear too tall. This is a typical example of the vertical elongation problem associated with the 4 / 3 aspect ratio in television signals and the unity aspect ratio in computer hardware, i.e., printers and monitors.
Consider a known shape, the circle, as a test pattern for displaying the vertical elongation problem associated with the television signal data obtained from VCR tapes. The circle, a theoretical shape in the following images will be considered computer generated and then video camera generated. The video camera generated image contains the 4 / 3 aspect ratio seen in signals from VCR crime scene tapes. The resulting images will be portrayed on the computer monitor, computer hardcopy printer, video monitor and video printer. The images are exaggerated to emphasize the vertical elongation problem due to the 4 / 3 aspect ratio inherent in television signals.
Figure 6 contains both the circle, the square, horizontal and vertical rectangles. The latter shapes are used to calibrate the digital magnification and subsequent scaling factors necessary to eliminate the vertical elongation problem in LaserJet printer hardcopy, the solution to the problem under discussion.
Figure 6
Test Images
Figure 6 Test Images
The following discussion deals with the various views of a circle and an oval, where the oval represents the television signal 4 / 3 aspect ratio or 133% vertical elongation of a circle. Note that the circle also displays a 3 / 4 aspect ratio horizontal elongation problem in certain environments.
Figure 7 Aspect Ratio Views of a Circle
Circle |
|
|
|
Video Printer |
||||
Inherent Aspect Ratio |
Unity |
3 / 4 |
Unity |
3 / 4 |
||||
|
||||||||
|
Circle |
Circle  Unity Aspect Ratio |
Horizontal Oval
|
Circle
|
Horizontal Oval
|
||||
|
Vertical Oval
|
Circle Unity Aspect Ratio |
Vertical Oval
|
Circle
|
||||
Figure 7 Aspect Ratio Views of a Circle
The various views will be discussed in individual groups.
The computer-generated circle appears as a circle on the computer
monitor and the computer printed hardcopy, Figure 8. Computer technology is a unity aspect
ratio world, no television aspect ratio appears in any of the images.
Figure 8 Views of a Computer Generated Circle on Computer Monitor - Laser Jet Hardcopy
|
Circle |
Computer |
LaserJet Hardcopy |
||
Inherent Aspect Ratio |
Unity |
Unity |
||
Circle |
Circle
|
Circle
|
||
Figure 8 Views of a Computer Generated Circle on Computer Monitor - Laser Jet Hardcopy
Figure 9 Views of a Video Generated Circle on Computer Monitor - Laser Jet Hardcopy
|
Circle |
Computer |
LaserJet
Hardcopy |
Inherent Aspect Ratio |
Unity |
Unity |
Circle |
Vertical Oval
|
Vertical Oval
|
Figure 9 Views of a Video Generated Circle on Computer Monitor - Laser Jet Hardcopy
The irony in all these aspect ratios is that the television camera generated circle appears as a circle on the television monitor but an oval on the computer monitor. The television camera has a 4 / 3 aspect ratio. The television monitor has a 3 / 4 aspect ratio. The application of these two aspect ratios in the world of television signals is unity aspect ratio. What you see is what you get in the television world. The problem arises when the television world and the computer world mix their signals.
Figure 10 Views of a Video Generated Circle on a Television Monitor and Video Printer
|
Circle |
Television |
Video
Printer |
Inherent Aspect Ratio |
3 / 4 |
3 / 4 |
|
||
|
Circle |
Circle
|
Circle
|
Figure 10 Views of a Video Generated Circle on a Television Monitor and Video Printer
The computer-generated circle appears as a oval on the television monitor and the video hardcopy, Figure 11. Television monitor technology is a 3 / 4 aspect ratio world. Televisions monitors by themselves display a horizontal elongation distortion of computer generated images. Together, the camera's 4 / 3 aspect ratio and the monitor’s 3 / 4 aspect ratio produce unity aspect ratio. A circle input to the camera appears as a circle on the television monitor, Figure 10.
Figure 11 Views of a Computer Generated Circle on a Television Monitor and Video Printer
|
Circle |
Television |
Video Printer |
||
Inherent Aspect Ratio |
3 / 4 |
3 / 4 |
||
|
||||
|
Circle |
Horizontal Oval
|
Horizontal Oval
3 / 4 Aspect |
||
Figure 11 Views of a Computer Generated Circle on a Television Monitor and Video Printer
The aspect ratio problem is evidenced by the ovals. Since the origin of
the data in Forensic Video Image Analysis is a television signal recorded on a magnetic
tape, the television camera input circle must appear as a circle on the television monitor
and on the LaserJet hardcopy. The various display formats as viewed in the
courtroom must be identical. This places a requirement on the LaserJet output to treat the
television signal in the same manner as the television monitor or the video printer and thereby apply an effective 3 / 4 Aspect Ratio to the computer printed images. This aspect
ratio adjustment applied to the LaserJet printer hardcopy can be calibrated from actual
images such as a square, rectangle, circle, an equilateral triangle or a pattern.
The VCR television format signal displayed on the computer monitor and
printed on LaserJet printer suffers the vertical elongation due to this aspect ratio. This
elongation can be eliminated by dynamically calibrated scale factors. Dynamic calibrated
scale factors allow unity aspect ratio computer printed images. Unity aspect ratio means
that what is seen in the courtroom, the crime scene video, the LaserJet hardcopy and the
video printer hardcopy are all the same. These images are all in agreement by the
application of the calibrated scale factors
In TREC’s Forensic Video Image Analysis software known as TRECWare, printing of analyzed images is undertaken using Adobe’s PageMaker program. The printing of a series of images is automated by a script file written during the analysis phase of the work. The script file is then transferred to the proper location, i.e., subdirectory, for access by PageMaker. The script file automatically loads each analyzed image, centers the image on the page in landscape layout format, slightly crops processing edges containing no information, labels the image with the filename, and sets up the first page with agency information, date, subdirectory information and video analyst initials. If the agency has a logo file, the logo is printed on the first page with the prior information on the second page. The images are then scaled to be identical with the images produced on the video printer
There is an economic benefit derived from this effort. The dye sublimation hardcopy images have a cost factor ranging from $1.75 to $2.50 per image. The LaserJet hardcopy images using the high resolution Hammermill Laser Plus long grain paper featuring wax holdout, model number 10450-5, at approximately $75 per case represents 3 to 4 cents per hardcopy image. This type paper avoids dot gain or ballooning upon fusing the graphics powder in the LaserJet process. It is hard on budgets to supply an investigator with 100 dye sublimation hardcopy images, but simply a matter of minutes to print 100 such sheets on a LaserJet, if the images are essentially identical.
The LaserJet used for Forensic Video Image Analysis is enhanced by XLI Corporation’s postscript engine from the initial 600 dots per inch to 2400 dots per inch. This halftone printing has a line screen frequency of 212, a bit less than the 300 line screen frequency specified for the continuous tone dye sublimation printers. Both printers provide 256 shades of gray. The LaserJet hard copies are admittedly a little less sharp, a bit lower resolution than the dye sublimation prints.
The following image along with the previous aspect ratio shapes was used for the dynamic calibration of the PageMaker script file scale factors. The image was obtained from a 35 mm slide created by Aperion, viewed by a Olympus model SZH zoom microscope at 7.5 magnification. The test pattern was read by a Panasonic CL 350 video camera and input directly into TRECWare, bypassing the time base correction unit. Note the vertical elongation distortion in what should be a symmetric image.
Figure 12 Test Pattern for Dynamic Calibrated Scale
Factors
Aspect Ratio Distorted by Television Camera
The scale factors required in PageMaker were measured using a high intensity diffused source light table with the LaserJet printed image overlaid on video dye sublimation printed hardcopy of Figure 12. Registration of the two overlaid images was within one line width or ± 2 pixels, on a line length of 512 pixels or ± 0.4%. The approach which produced the best comparison
of the two different printed images was to horizontally scale the laser jet printer image approximately 10% both left and right. The actual scale factor was measured to be 10.15%. TREC’s Forensic Video Image Analysis software, TRECWare, the PageMaker printer produces the same results as the video printer, with slightly less resolution, i.e., near photographic quality.Conclusion Down Load Zipped Word Doc File
Therefore, what you see in the courtroom, on the television monitor portraying the crime scene video, on the dye sublimation hardcopy and on the LaserJet hardcopy are all the same. The problem of vertical elongation or aspect ratio distortion in LaserJet printed images from crime scene videotapes is solved.
