Digital Rectification and Resizing Correction of Photographic Bite Mark Evidence by Bowers and Johansen (Forensic Science Communications, July 2001)
July 2001 - Volume 3 - Number 3
Research and Technology
Digital Rectification and Resizing Correction of
Photographic Bite Mark Evidence
C. Michael Bowers
Deputy Medical Examiner
Ventura Medical Examiner's Office
Raymond J. Johansen
Forensic Dentistry Consultant
Santa Barbara Sheriff's Office
Santa Barbara, California
Bite mark analysis has been judicially accepted in the United States since 1954 (Doyle v. State). The identification of a specific biter has been instrumental in criminal investigations of homicide, sexual assault, and child abuse cases (Pretty and Sweet 2000). The majority of bite mark cases involve photographs of bite marks on skin and other substances that are later associated with known dental evidence obtained from suspects (Krauss 1984). This comparative analysis primarily uses superimposition of these evidence samples. Therefore, the dimensional accuracy and sizing of both evidence images are of utmost importance. Forensic protocols for the photographic reproduction of crime scene evidence demand that a linear scale be placed next to the evidence sample to make an accurate comparison (Hyzer and Krauss 1988). This known dimensional reference allows the photographic examiner to re-create life-sized graphical reproductions. The presence of photographic distortion is evidenced by the scale's incremental lines appearing nonparallel and not uniformly shaped (Bernstein 1985). Without rectification, the photographed evidence sample will not be representative of its true shape and dimension.
The American Board of Forensic Odontology allows digital imaging in bite mark analysis (Bowers and Bell 2001). The correction of photographic distortion is usually the realm of the photographic laboratory technician.
|Figure 1. A bite mark on skin that is vague, diffuse, and incomplete. The paper scale shows nonparallel incremental lines and surface undulations. The scale's dimensions should be one inch square. Each incremental square is one-tenth of an inch. Click for enlarged image.|
|Figure 2. Close-up of the G scale. It is nonrigid and shows undulations. The distortions of the outer inch square grid and one-tenth square rectangles indicate the entire evidence image cannot be reproduced accurately to life-sized (1:1) until the distortion correction is made. Click for enlarged image.|
|Figure 3. Original G-scale image. The internal angles should all be 90 degrees. The scale leg marked by the yellow double-headed arrow is 7 percent shorter than the adjacent black arrow. Click for enlarged image.|
|Figure 4. The vertical (blue) and horizontal (yellow) guides are placed along edges of the G scale. The orange arrows indicate the portions of the scale used to resize the height (five-tenths of an inch) and width (four-tenths of an inch) of the entire evidence image to life-sized (1:1). Using this portion of the scale to resize the entire evidence image is possible if the bite mark is on the same plane as the scale. Click for enlarged image.|
|Figure 5. Info palette showing the relative angle (A) and the measuring line's length (D).|
|Figure 6. Rotate canvas dialog box displays the angle of misalignment before digital rectification.|
|Figure 7. The G scale and the blue guides are in proper alignment. Click for enlarged image.|
|Figure 8. Info palette records original height and width dimension (D) taken with the measure tool before digitally resizing the image.|
|Figure 9. The injury pattern is now an accurate life-sized reproduction by use of the 0.5 inch and 0.4 inch legs of the G as a dimensional reference. Click for enlarged image.|
|Figure 10. Teeth edges of the suspect's dentition developed from a digital scan of plaster dental models and image editing tools of Adobe® Photoshop®. Click for enlarged image.|
|Figure 11. Edges of the defendant's teeth placed next to the corrected image of the bite mark. The forensic value of this particular comparison is negligible because of the poor definition of the original injury pattern. The prosecution entered the original uncorrected bite mark photograph into evidence and argued that a positive dental identification of the defendant had occurred. Click for enlarged image.|
Johansen and Bowers (2000) and Sweet and Bowers (1998) tested and used the digital imaging program Adobe® Photoshop® (Adobe Systems Incorporated, San Jose, California) for its adjunctive value to photographic evidence evaluation and comparative analysis. It was used in this case to correct for certain types of distortion created from misplacement of the linear reference scale (parallax distortion) and off-angle placement (perspective distortion) of the camera.
This article is, therefore, a description of methods recently used in the analysis of bite mark evidence from a 1973 homicide that was prosecuted in 2000 (Figure 1).
The only photograph of the bite mark injury on skin was a black-and-white print that was distorted because of a combination of camera misalignment and discrepancies in the shape of the scale placed next to the injury (Figure 2).
The dimensional accuracy and forensic identification value of this photographic evidence were evaluated. Digital rectification methods were used to correct the shape distortion and render the photograph amenable to later comparison with a defendant's teeth by digital superimposition.
This protocol uses the digital editing program Adobe® Photoshop® 5.5, a desktop computer, and a flatbed scanner. Photoshop® is a commercially available digital imaging program that runs on laptop and desktop computers having sufficient chip speed and RAM (Pentium II or equivalent and at least 96 MB of RAM) to open the large files generated by high-resolution digital capture devices.
The original evidence photograph was scanned at 300 dpi on a Microtek Scanmaker V6USL flatbed scanner (Microtek International Incorporated, Redondo Beach, California). It was then imported into the Adobe® Photoshop® 5.5 editing program. A Dell™ Dimension XPS T500 computer with a 20-GB hard drive and 256 MB of RAM was used for this analysis.
After it was imported into Photoshop®, the original image was adjusted for contrast, rotated to orient the horizontal leg of the scale along the computer screen's x-axis, and cropped.
The sides and internal angles of the G scale were measured using the Photoshop® measure tool (Figure 3).
Portions of the G scale within the image were then returned to their original shape using the Photoshop® distort function (Figure 4). The steps involved are outlined below (Johansen and Bowers 2000, pp. 18–41).
- Select the measure tool from the toolbox, and draw a line along the horizontal or vertical edge of the scale (Figure 4). Note that in the Photoshop® info palette (Figure 5), the A displays the angle drawn relative to the zero (x) or 90 (y) axis. The D represents the distance or length of the measuring line and is unimportant for this step.
- Next, starting from the top menu bar, click Image > Rotate Canvas > Arbitrary.
- The rotate canvas dialog box (Figure 6) appears and displays the angle that the measure tool is off either along the horizontal or vertical axis.
- This represents the amount of rotation necessary to bring the measuring line parallel to the x- or y-axis. In this case, the x-axis was chosen. Photoshop® indicates in this dialog box the amount of angular correction necessary to level the scale's leg onto the zero (x) axis. Click OK to rotate the image 11.36 degrees clockwise (Figure 4).
- The image was then cropped to eliminate unnecessary peripheral areas.
- Guides were brought onto the image and aligned along the left vertical and lower horizontal edge of the G scale. The entire image was selected.
- The distort function was repetitively used to align the indicated sides of the scale with the blue and yellow guides (Figure 7).
The dimensions of the G scale were not life-sized, so the image was resized to (1:1). Undistorted portions of the vertical side and horizontal legs of the scale closest to the bite mark were used to resize the entire image.
A brief description of the steps involved follows:
- The measure tool is used to measure the G scale's undistorted horizontal and vertical legs. The actual height is 0.5 inches, and the actual width is 0.4 inches.
- The info palette (Figure 8) records the dimension of these measurements (D). The 0.30 (inch) value indicates the horizontal component of the scale's image is smaller than life-sized. A indicates this measurement is on the zero (x) axis of the computer screen.
The image dimensions (height and width) were separately changed by the ratio of the actual dimensions of the accurate portions of the scale to the dimensions existing in the info palette (height ratio was 0.5/0.33, width ratio was 0.4/0.3; Johansen and Bowers 2000, p. 46).
The completed resized image was determined to be an accurate life-sized representation of the bite mark evidence when the selected portions of the G scale have proper shape and dimension. Placing digital grids of one-tenth inch dimension over the G scale and visually determining acceptable correlation checks this.
The internal angles of the G scale were measured after rectification and resizing (Figure 9). The nondistorted portions of the vertical and horizontal legs of the G scale were also measured to evaluate the success of the resize procedures. The results indicate that the portions of the digitally remastered G scale have proper dimensions and shape proportions.
Subsequent digital superimposition of the defendant's dentition (Figure 10) was performed over the bite mark image (Figure 11).
The use of the described digital methods is amenable to evidence photographs that contain a scale of known dimension. The two-dimensional (an x- and y-axis) L-shaped ABFO No. 2 scale (Lightning Powder Company, Incorporated, Salem, Oregon) allows the examiner to independently resize the width and height of the image. The presence of a one-dimensional scale (ruler) requires a one-step resize when the length of the scale is used to simultaneously re-dimension the width and height of the evidence image. In this case, the G scale was used as a two-dimensional reference.
ABFO bite mark methodology guidelines. In: Manual of Forensic Odontology. C. M. Bowers and G. L. Bell, eds. American Society of Forensic Odontology, Colorado Springs, Colorado, 3rd ed. 2001, p.335.
Bernstein, M. L. Two bite mark cases with inadequate scale references, Journal of Forensic Sciences (1985) 30(3):958–964.
Doyle v. State, 159 Tex.Crim310, 263 S.W.2d 779 (App.1954).
Hyzer, W. G. and Krauss, T. C. The bite mark standard reference scale—ABFO No. 2, Journal of Forensic Sciences (1988) 33(2):498–506.
Johansen, R. J. and Bowers, C. M. Digital Analysis of Bite Mark Evidence. Forensic Imaging Services, Santa Barbara, California, 2000.
Krauss, T. C. Photographic techniques of concern in metric bite mark analysis, Journal of Forensic Sciences (1984) 29(2):633–638.
Pretty, I. A. and Sweet, D. Anatomical location of bitemarks and associated findings in 101 cases from the United States, Journal of Forensic Sciences (2000) 45(4):812–814.
Sweet, D. and Bowers, C. M. Accuracy of bite mark overlays: A comparison of five common methods to produce exemplars from a suspect's dentition, Journal of Forensic Sciences (1998) 43(2):362–367.