Blow It Up
Written by David Menor Witzke   


One of the earliest milestones in the history of using digital imaging technologies to present fingerprint evidence in court occurred more than 25 years ago. The first case to establish a precedence for the acceptance of digitally enhanced evidence was the Commonwealth of Virginia v. Robert Douglas Knight (1991). The trial of a March 1990 murder case involved the enhancement of a bloody fingerprint found on a pillowcase at the crime scene. During the trial, the defense attacked the image processing used to enhance the print, and requested a Kelly/Frye hearing to determine the scientific validity and acceptance of the digital enhancement process. The court determined that the techniques were, in essence, consistent with traditional photographic processes, and Robert Knight pleaded guilty and was sentenced to four life terms.
Ten years later, in the State of Florida v. Victor Reyes (2001), the defense challenged the use of certain digital image processing practices for the enhancement of fingerprints that had been developed on duct tape. In particular, the defense challenged the use of Adobe Photoshop to adjust contrast and suppress background noise. The court found that “the use of digital imaging to enhance a latent print is not new or novel and that it is accepted within the relevant forensic community,” and that “the process of recording of digital enhancement processes is actually more thorough than the recording process of traditional photographic processes…”.
Reyes is a milestone because there was a major advancement that came out of that case: the bubble chart (Figure 1). To support my testimony during the Frye hearing, I developed what is now commonly known as a “bubble chart” to not only help demonstrate the ACE-V process, but to demonstrate that the analysis and comparison process is more than just counting a series of “points”.

Figure 1 — Bubble charts are now used frequently in the latent print community to document the comparison process in countries such as the United States, Canada, Europe, Australia, and New Zealand. In fact, this technique has been copied in a number of AFIS solutions, such as AFIX Tracker, and other analysis and comparison tools, such as CSIPix, FCS, etc.
However, preparing and presenting this level of documentation requires more than a basic understanding about resolution. The examiner must also understand how resolution affects image quality, from the point of acquiring the original image to preparing the documentation for testifying in court. It is also crucial to understand how digital imaging technologies can be employed to assist with making a point in court when explaining the process that was used for comparison—for example, enlarging the images so that we can view and explain Level 1, Level 2, and Level 3 detail more clearly and easily. It also provides the ability to demonstrate how we count and use ridge structures to compare the continuity in the location of ridge events using ridge counts from one event to another. And, most importantly, it helps demonstrate how Level 3 detail (such as ridge shape and direction, pores, scars, creases, deformations, and incipient ridges) correspond from one impression to the next.
Image resolution, which is the prime factor in image quality, always starts with the device used to capture the digital image. For example, using a flatbed scanner, the number of pixels increase in direct proportion to the size of the area scanned times the number of pixels specified (misidentified in the scanner interface as DPI) and the number of pixels in width times the total number of pixels in height. For example, a 1-in. area scanned at 1200 PPI would create an image with 1,440,000 px (i.e., 1200 x 1200 px) with a file size of approximately 4.12 MB. A 2-by-3-in. area scanned at 1200 PPI would create an image measuring 2400 x 3600 px, for a total of 8,640,000 pixels, with a file size of approximately 24.72 MB). A 3-x-5-in. area scanned at 1200 PPI would create an image measuring 3600 x 6000 px, for a total of 21,600,000 pixels (with a file size of approximately 61.8 MB).
Since both the size of the area and the resolution is set before the image is scanned, images acquired from a flatbed scanner do not require calibration. There is no depth of field from the object to the face of the scanner, so the size of the area is already for 1:1 (life-size) capture.
The opposite effect occurs using a digital single-lens reflex camera (also called a digital SLR or DSLR camera). The imaging sensors in a digital camera have a fixed number of photo receptors, and vary not only from manufacturer to manufacturer, but from model to model. Table 1 (below) is provided for illustrative purposes only, and does not represent the wide range of imaging sensors available today.
Number of Photo Receptors (Width) Number of Photo Receptors (Height) Total Number of Photo Receptors (Width x Height) Number of Megapixels


3,262 16,075,136 16.1
4,992 3,328 16,613,376 16.6
5,184 3,456 17,915,904 17.9
5,616 3,744 21,026,304 21.0
6,048 4,032 24,385,536 24.4
7,360 4,912 36,152,320 36.2

Table 1—Camera manufacturers typically use image sensors that are different sizes and with different numbers of megapixels. This table is provided only to illustrate the relationship between the number of pixels (width and height), total number of image pixels, and number of megapixels.

Using a digital camera with a resolution of 24.4 MP, capturing an area (such as a footprint) that is approximately 12 x 8 in. wide would result in a resolution of approximately 500 PPI. Capturing an area that is approximately 6 x 8 in. would result in a resolution of approximately 1000 PPI. Capturing an area (such as a palm) that is approximately 4 x 2.7 in. would result in a resolution of approximately 1512 PPI. Capturing an area (such as a single fingerprint) that is approximately 2 x 1.3 in. would result in a resolution of approximately 3024 PPI.
The bottom line is that not only must a digital image captured using a digital camera be calibrated, but a digital camera with the appropriate resolution must be available to capture items at the required resolution (i.e., latent impressions—fingerprints and palmprints—must be captured with a minimum of 1000 PPI).
Using the appropriate lens, such as a macro lens (60mm macro lens for Nikon; 50mm macro lens for Canon) with a full-frame (36 x 24 mm) imaging sensor, is also a requirement when photographing evidentiary images. In accordance with best practices, a minimum of two photographs should always be captured of the evidence, even in the laboratory. One photograph should be an “overall” photograph that illustrates placement and position of the impression; and the second photograph should be captured with the lens as close to the object as possible for the highest possible image resolution and the best possible image quality.
Once the image has been captured at its highest possible resolution, the original image must be preserved for purposes of authentication. All processing should be done on a copy of the original image.
While most latent print examiners and forensic photographers know not to change the resolution of an image unless it is required for a specific purpose, such as to send to an automated fingerprint identification system (AFIS), they often forget that rotating an image does not change the calibrated resolution, but it will affect the overall size of an image as well as physically change the original pixel values throughout the entire image, as shown in Figure 2.
Figure 2—From left to right: The vertical red (R), green (G), and blue (B) lines are each 1 pixel wide, for a row of 3 pixel values. When the lines are rotated at 22.5 degrees, the row of 3 pixels increase to alternating rows of 4 and 5 pixels; when rotated at 45 degrees, the row of 3 pixels increase to alternating rows of 6 and 7 pixels; when rotated at 75 degrees, the row of three pixels expand to alternating rows of 15 and 16 pixels.
Rotating images is not the only process that creates significant aliasing artifacts. Resampling (upsampling or downsampling) occurs when changing image size as well as when sending a digital image to a printer. Downsampling decreases the original number of pixels in an image (discarding actual pixel values in the image), such as when a picture is reduced in size to produce a 1:1 printout; conversely, upsampling increases in the number of pixels (creating pixel values that did not previously exist within the image), such as when a picture is increased in size so a small image can be printed as a large picture.
Resizing an image is not simply a one to one process, it is a one to x2 process. For example, imagine having a 1 x 1-in. square box: the box would cover an area that is 1 sq. in. If you were to increase the size of this box to 2 x 2 in., the box would now cover 4 sq. in. The enlargement would result in an area with 3 additional square inches, or an increase of 300%, as illustrated in Figure 3.

Figure 3—Making a 2x enlargement of a square that is 1 x 1-in. would result in a square that is 4 sq. in. for an increase of 300% (red square); a 3x enlargement of the 1 x 1-in. square would result in an area that is 9 sq. in., for an increase of 800% (blue square). Continuing with that, a 6x enlargement would result in an area that is 36 sq. in., for an increase of 3500% (purple square).
Using these dimensions in terms of pixels:
  • A 1-in.-square image with 500 pixels per inch (PPI) would be 500 px wide by 500 px high, for a total of 250,000 px.
  • A 2-in.-square image with 500 PPI would be 500 + 500 px wide for a total of 1000 pixels wide, and would be 500 + 500 px high for a total of 1000 pixels high, thus having a total of 1,000,000 pixels.
  • A 3-in. square image with 500 PPI would be 500 + 500 + 500 px wide for a total of 1500 px wide, and would be 500 + 500 + 500 px high for a total of 1500 px high, thus having a total of 2,250,000 pixels.
  • A 4-in. square image with 500 PPI would be 500 + 500 + 500 + 500 px wide for a total of 2000 px wide, and would be 500 + 500 + 500 + 500 px high for a total of 2000 px high, thus having a total of 4,000,000 px.
Downsizing (a.k.a. resampling) has the absolute opposite effect on digital images. For example, resampling an image from 1000 PPI to 500 PPI is a reduction of pixel information… in this case, a reduction of 300 percent of the original pixel values using the same 22:1 ratio. Even resampling a 1 x 1-in. image scanned at 1200 PPI to a resolution of 1000 PPI results in a loss of 440,000 pixel values, or a decrease of 44% of the original pixel values/image quality, as illustrated in Figure 4. The resulting loss of data can have a significant impact on how image data is analyzed and compared. While changing image resolution cannot always be avoided, it is recommended that the image with its full (unchanged) resolution be retained, and the resampled image be used only as a temporary image, and then discarded after its use.

Figure 4—It is challenging to illustrate the visual difference in a digital image with a calibrated resolution of 2146 PPI (left) downsampled to 1000 PPI (right) in a medium other than on-screen, but the pixilation is more pronounced in the image on the right. It is important to remember that while the human eye may not notice the distinction between the pixels, the computer does notice and uses every single pixel value during image processing.
All digital images from devices such as digital cameras, flatbed scanners, and digital video cameras, etc. are based on a format commonly known as a bit-mapped graphic (a.k.a. bitmap), which consists of a square grid of pixels and is commonly referred to as a “raster graphics image”. Bitmapped (pixel) graphics cannot be enlarged beyond a certain size without degradation (pixilation) depending upon the image resolution. The higher the resolution of the image, the bigger the image can be enlarged. All computer monitors, LCD projectors, and printers are also raster devices.
Drawings that are created on top of digital images, such as annotations, are based on the use of vector graphics. Vector graphics are not made up of pixels, but are created using “points” and geometric shapes to create shapes or paths, such as a line drawing. All vector graphics must be “rasterized” (converted to a pixel format) before they can be displayed or printed. Some software applications and printers employ the use of a tool (driver) known as a Raster Image Processor (RIP) to convert vector graphic images into a raster format for printing as well as enlarging raster (pixel) format images to minimize degradation.
It is easy to convert an image file from a vector format to a bitmap or raster file format, but it is much more difficult to go in the opposite direction, especially if the underlying image is a raster format from a digital camera, flatbed scanner, or video camera, since these devices produce images that are continuous-tone raster (pixel) graphics that are impractical to convert into vectors. Most, if not all, image-editing applications (including Adobe Photoshop) must operate on the individual pixel values that actually make up the digital image file. These applications also provide vector-based tools for creating graphic overlays on top of the raster image, such as for the preparation of court exhibits, that must be rasterized before the image can be printed.
Even individual vector graphic layers must be rasterized before adding additional detail to a specific layer. For example, in Adobe Photoshop, create a new layer with text. Then, using the line tool, try to create a line from an artifact within the image to the text on the same layer. An error message similar to the one shown in Figure 5 will be displayed.

Figure 5—Text layers, which use vector graphics, must be rasterized before adding lines or other graphics on that layer.
It should also be noted that users typically must save images created from a vector source file as a raster format because different systems (monitors, printers, and even software applications such as Adobe Photoshop) have different (and often incompatible) vector formats, if they even support vector graphics at all.
It is also pointless to use compressed digital images (images that have been compressed using a lossy JPEG compression algorithm) in conjunction with vector-based graphics. When a compressed digital image is enlarged, the artifacts created by the JPEG compression algorithm can appear quite obvious… even if the graphic overlay does not appear degraded. For example, while the overlaying graphics (such as the lines and letters or numbers) used in the court exhibit may not appear to be degraded when the image is enlarged, the artifacts created during the compression process in the underlying pixel-based digital image can be very apparent. In addition, low-resolution digital images do not enlarge very well either, so even if the overlaying graphics can be enlarged the underlying digital image will still appear.
Even with an understanding of image quality (resolution) and image processing, there remains the question of how to analyze and compare the image(s). Will the image be displayed on a computer monitor? What is the size and resolution of that monitor? Or, will the image be printed for analysis and comparison? Will the image be printed on an ink jet printer, a laser printer, or a dye sublimation printer? Or, will the image be sent to AFIS?
Today, many forensics experts—including footwear and tire tread examiners, firearms examiners, latent print examiners, and questioned document examiners—know and understand that output technology has lagged far behind digital image capture/digital display technologies, yet they still want to use a 1:1, life-size, printout of the digital image.
While viewing an image on-screen does not physically change the actual number of image pixels, the percent of magnification (zoom rate) can affect the level of detail visible on the display. However, sending an image to a printer can degrade the image quality substantially because images must be resampled to the resolution of the output device, and actual pixel values are discarded. Adobe Photoshop provides a number of different resampling methods, including:
  • Automatic—a method of resampling where Photoshop chooses the method of resampling based on the contents (pixel values) of an image and whether pixel values are being added to or being removed from the image;
  • Preserve Details—a resampling method, like Bicubic Smoother, that is intended for enlarging images. However, with this method, a Noise reduction slider is used for reducing the noise as additional (new) pixels are added to the image.
  • Bicubic Smoother—a resampling method primarily intended for enlarging images where the added pixel values are based on surrounding pixels to produce pixel values with smoother tonal gradients for the added pixels.
  • Bicubic Sharper—a resampling method primarily intended for reducing the number of pixels where the remaining pixel values are based on the original surrounding pixel values and then sharpened to maintain contrast.
  • Bicubic—a more precise method of adding or removing pixel values based on surrounding pixel values to produce smoother tonal gradations without edges than Nearest Neighbor or Bilinear.
  • Nearest Neighbor—resampling based on pixel values where non-anti-aliased hard edges are preserved.
  • Bilinear—resampling that adds or removes pixel values based on averaging the pixel values of surrounding pixels, producing medium-quality results.
One method is not always the best solution for all types of images, as illustrated in Figures 6, 7, and 8.

Figure 6—(Left to right) Original image; image upsampled from 500 to 1000 PPI using “nearest neighbor” resampling and the details are as clear as the original values; image upsampled from 500 to 1000 PPI using “preserve details” resampling, where the details are not preserved very accurately.

Figure 7—(Left to right) Original image; image upsampled from 500 PPI to 1000 PPI using “bicubic” resampling; image upsampled from 500 to 1000 PPI using “bilinear” resampling. In both cases, the white space between the lines becomes a gray value (based on the averaging of the original black and white pixel values), and the lines lose their sharpness.

Figure 8—(Left to right) Original image; image downsampled from 1000 to 300 PPI using “nearest neighbor” resampling; image downsampled from 1000 to 600 PPI using “nearest neighbor” resampling. The higher the rate of resampling, the more detail is lost within the image. In the case of latent prints, Level 3 detail can be lost very easily.
Most, if not all, evidentiary images (calibrated 1:1) must be downsampled for printing, in which actual pixel values are discarded and the remaining pixel values go through a process called “dithering,” where the pixel values are converted to a “cluster of dots.” Dithering creates an approximate color interpretation of the image pixels using a combination of the ink colors and dot sizes available in the printer.
NOTE: The type of printer, the printer settings, and the type of paper used has a substantial impact on image output quality, and varies from printer to printer.
The process of converting pixel values to dots is probably one of the most grossly misunderstood processes within the digital imaging process. For example, walk into any store that specializes in the sale of digital technologies (i.e., digital cameras, flatbed scanners, and printers), and ask them how many pixels are printed using a particular laser or ink jet printer. The answers that you receive will not only amaze you, but will be good conversation starters at your next forensic group meeting.
If you really want to have fun, ask the sales person how many different color values can be printed by the printer. All color laser printers have only four colors of toner: cyan (C), magenta (M), yellow (Y) and black (K), or CMYK. Since all digital imaging devices (digital cameras, flatbed scanners, video cameras, computer monitors, and LED screens) have red (R), green (G), and blue (B), or RGB, how does a laser printer print a red, green, or blue pixel value? A knowledgeable sales person would say that it takes a cluster (group) of dots to “give the appearance” of a single pixel value (Figure 9); the untrained sales person will most likely tell you that the terms dots and pixels are interchangeable, and the inks are mixed together in the printer to create the pixel value, therefore, the number of dots and pixels are the same.

Figure 9—Most black-and-white laser printers use a four-dot pitch (4 x 4) grid to produce different shades of gray. Ink-jet printers use multiple ink colors of varying sized droplets placed closely together to “trick” the human eye to see a specific pixel color value. Ink-jet printers with a high number of droplets per inch, with multiple colors of ink coupled with small droplet sizes (starting at 1.5 picoliters) will produce the most accurate, reliable output.
Therefore, the larger issue is not only how do you explain the effect of the digital imaging processes on the image, but how do you explain the effect of printing on the quality of the image output?
I recently had the opportunity to review the output from a Noritsu QSS Green II Duplex Inkjet System at the South Carolina Law Enforcement Division (SLED) in Columbia, South Carolina. While I have not had the opportunity to conduct extensive testing of this device, the output was astonishing… and welcome from a forensic point of view. For the first time ever, I was able to send a digital image that contained 1000 pixels (with a resolution of 1000 PPI), and the printer was able to recreate all 1000 pixel values.
The 1000 PPI test print consisted of series of 10 lines that were 100% black and 1 pixel wide, where each line was separated by a white space that was also 1 pixel wide. While the dithering of the CMYK dots in the white space did not provide a perfectly white value, the output did, however, accurately reproduce each of the 1-pixel lines, as shown in Figure 10.
Figure 10
Figure 11
Figure 10 & 11—In each pair of examples, the original test target is displayed on the left, and the output from the Noritsu QSS Green II Duplex Inkjet System is shown on the right. The printed image was scanned at 4800 PPI and then enlarged to match the enlargement of the test target for comparison.
Another part of the test target consisted of a series of 19 lines that were 1 pixel wide, where each line was separated by a white space that was also 1 pixel wide. However, this series of lines was used to test the printer’s ability to reproduce grayscale values. The grayscale values started at 10% grayscale, and increased in increments of 5%.
As shown in Figure 11, the output values for the top three lines (10%, 15%, and 20% grayscale) are almost indistinguishable. Similarly, the dot patterns for the next two lines (25% and 30% grayscale) are nearly identical. As a result, it would be almost impossible to discern the difference between latent impressions with minimal contrast between light (faint) ridges and adjacent furrows. It should also be noted that this test was conducted using only 19 shades of gray, whereas standard, 8-bit digital images are comprised of 256 different shades of gray.
Figure 12 illustrates the output from a test target that was created by the MITRE Corporation and has been used by the FBI to measure output quality of a number of output devices. The test target was created at 500 PPI (based on the previous AFIS standard of 500 PPI). Similar to the target shown in Figure 11, the 5 black lines in the target in Figure 12 are 1 pixel wide and the white space separating the lines are also 1 pixel wide. The various outputs in Figure 12 were scanned at 4800 PPI and then enlarged to the same magnification as the test target for 1:1 comparison.

Figure 12—From left to right, top row: 1) the original test target, 2) output from a Kodak 8660 Dye Sublimation printer, and 3) output from a Fargo DTC Dye Sublimation printer. Bottom row: 4) output from an Epson Stylus Photo R2000 Inkjet Printer, 5) output from a Fujifilm ASK-4000 Dye Sublimation printer, and 6) output from an HP Laser Jet printer.
If there was ever an argument for doing on-screen analysis and comparison, this is it. Analysis and comparison of latent prints, palm prints, footwear, tire tread, firearms, tool marks, blood spatter, trace evidence, etc., can be done much more accurately, effectively, and efficiently using a high quality, high resolution LED monitor. All computer monitors used by forensic experts display a significantly higher number of pixels than a printer can print, and more and more forensic examiners are relying on high-resolution monitors for analysis and comparison. In addition, they can zoom in and zoom out to take advantage of every single pixel value contained within an image, since the image detail is not physically lost as is the case with printing. On-screen analysis and comparison can provide a greater level of accuracy, especially when looking for Level 3 fingerprint detail such as the size and shape of pores, size and shape of ridge edges, incipient ridges, and so forth.
Helpful Hint: I recommend the LG 34UC97 Cineview Curved 34-inch Ultrawide LED Monitor, which has a resolution of 3440 x 1440 px (4.95 MP). Most high-resolution 24- and 27-in. monitors provide a maximum resolution of only 1920 x 1080 px (2.07 MP); even with dual monitors, the maximum resolution would be 3840 x 1080 px. When viewed at a 100% zoom, the 1440 pixels allow you to view almost an entire 11-inch page, and two pages at 8.5 in. wide easily fit on the screen side-by-side.
On the topic of on-screen analysis and comparison, I developed what has commonly been referred to as a “bubble chart” to explain the analysis and comparison process in the Reyes case. This technique is now widely used throughout the law enforcement community in the United States, Canada, Europe, Australia, New Zealand, etc. In fact, this technique has been copied in a number AFIS solutions, such as AFIX Tracker, and other analysis and comparison tools, such as CSIpix and FCS.
However, I believe generally that the forensic science community is not fulfilling scientific methods or principles when conducting impression analysis and comparisons visually. The “scientific method” is defined based on four steps:
1) Observation and description of a phenomenon or group of phenomena.
2) Formulation of an hypothesis to explain the phenomena... often takes the form of... a mathematical relation.
3) Use of the hypothesis to predict the existence of other phenomena, or to predict quantitatively the results of new observations.
4) Performance of experimental tests of the predictions by several independent experimenters and properly performed experiments.
When analyzing and/or comparing impression evidence visually, an accurate and reliable evaluation of the mathematical relationship of a digital image can be severely impaired when the original pixels are reduced from the original resolution to 1000 PPI. In many instances, the 1000 PPI resolution is further degraded using printer drivers that reduce not only the number of pixel values, but also change the color profile of the original image from RGB to a CMYK output profile. As stated earlier, a simple 8-bit grayscale image can include up to 256 different shades of gray whereas even the most expensive, high-end printers can produce a limited number of “distinguishable” shades of gray. In fact, most commercially available laser printers can produce only about 17 shades of gray, while color ink jet printers can produce slightly more than 50 shades of gray. The bottom line is that detailed image analysis (Level 2 and Level 3 detail) should not be performed on an image where the number of pixels and/or the original pixel values have been changed or lost in their entirety.
The good news is that today’s technology provides a far more superior and robust image analysis and comparison process. More importantly, the issues of physically changing the number of pixels and pixel color values is avoided in their entirety!
Just a couple of years ago, the forensic science community was using monitors with a resolution of 1024 x 768. Although there are still a number of agencies that still use these types of monitors, more law enforcement agencies are upgrading to 24- or 27-in.-wide monitors with resolutions of 1920 x 1080 px, 1600 x 1024 px, or 1280 x 800 px.
Monitors with high resolution allow full-screen display of calibrated, digital images; however, it is difficult to compare two images that each have a width of 1200 pixels (for a total of 2400 pixels) using a monitor that can display only a total of 1024 pixels. (You would have to zoom out and view less than 50 percent of the image pixel values.) Using the latest wide-screen monitor technology, you can compare images side-by-side with resolutions up to 3440 x 1440 px or 2560 x 1440 px, with image color values in excess of 16.7 million different color values.
While it has not yet happened in the forensic science community, the courts are becoming much more savvy about technology and how that technology can and should be used. In a New Jersey wrongful death/malpractice case, Rodd v. Raritan Radiologic Associates, P.A., 860 A.2d 1003 (N.J. Super. App. Div. 2004), the defendant (a radiologist) used a magnifying glass to examine the x-ray images from the decedent’s 1997 and 1998 mammograms and found them negative for cancer.
During the trial, the defense stated that mammograms were studied to detect cancer by placing the x-ray in a view box and analyzing it with his naked eye and a hand-held magnifying glass.
To help the jurors understand the finer details of the x-rays, the plaintiff digitized and “super-magnified” portions of the contested x-rays, and displayed them in court on a 6-x-8-ft. projection screen.
In response to numerous objections by the defense, the trial judge reasoned that the enlarged x-ray images not only helped the jurors understand the complicated factual scenario, but he also believed that the use of this process would be a meaningful improvement to the entire radiological field:
“The message may get—get out now, that in radiology, and I know the radiologist is under awesome pressure reading these films, that maybe they ought to blow it up like that… [M]aybe the whole industry is negligent. Maybe in this case, something ought to be done… maybe this message is gonna get out… it seems, to me, to be very simple and very easy to implement, in a radiology group, blowing it up on a screen… (emphasis added)”
The defense argued that the “standard of care” among radiologists required only a visual inspection with a 2.5x magnifying lens. Not only did the appellate court uphold the trial judge’s reasoning in admitting the digitized and “super-magnified” portions of the contested x-rays, but went on to highlight the fact that the plaintiff’s malpractice claim focused on an “error in visual observation.” (After deliberations, the jury returned a verdict awarding the plaintiff $1.5 million for loss of consortium; $1.7 million in survivorship damages; and $40,000 for wrongful death.)
There are, however, other issues commonly referred to as “The Swinton Six” that were called into question in this case, including, but not limited to:
1) The computer equipment is accepted in the field as standard;
2) Proper procedures were followed;
3) A reliable software application was used;
4) A qualified operator was used to create the exhibits;
5) Computer-generated exhibits must be proven to be a “faithful representation of the subject at the time in question”; and
6) The processes used were reliable and can be fully explained using the testimony of a knowledgeable witness who can be examined and cross-examined about the functioning of the computer as well as the processes used to create the exhibit.
Digital imaging provides the ability to dramatically improve the quality of forensic evidence for more accurate and reliable analysis and comparison. The proper use of image analysis and comparison tools also considerably increases the ability to document the analysis, comparison, evaluation, and verification (ACE-V) process easier and more effectively that using older (primitive) and supposedly “simpler” methods, such as using printed 1:1 (life-size) printouts for comparison even though, technologically, digital image processing provides faster, more efficient analysis and comparison of images using high-quality, high-resolution monitors.
But even with all of the tools, features, and functions that are available, starting with a high-resolution image will result in a higher-quality digital image that provides better visibility of image detail.
While fingerprint experts have presented digital “evidentiary” images in courts since the mid 1990s, the examiners’ knowledge and skill of fingerprint identification are being challenged. But with the right tools, the examiners can clearly demonstrate how the identification process is consistent with current scientific and scholarly knowledge.
About the Author
David “Ski” Witzke is the vice president of program management for Foray Technologies. He has more than 20 years of AFIS and forensic digital imaging experience and has conducted hundreds of digital imaging training programs for law enforcement agencies throughout the U.S. and Canada.
Azcarate, P. “Digital Imaging: The Technology and the Prosecutor,” The Prosecutor Magazine (Jan/Feb 2000, Volume 34, Number 1).
Berg, E. “The Digital Future of Investigations,” Law Enforcement Technology (August 1995).

Berg, E. “Legal Ramifications of Digital Imaging in Law Enforcement,” Forensic Science Communications, FBI Laboratory Services (Jan/Feb 2000, Volume 2, Number 4.

Davidson, M. W. and the Florida State University. “Basic Concepts in Digital Image Processing,” Optical Microscopy Primer: Digital Imaging in Optical Microscopy. Retrieved August 15, 2016 from:

Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 US, 579 (1993).

Federal Rules of Evidence, Article X. Contents of Writings, Recordings, and Photographs, Rules 1001(1), 1001(3), 1001(4), 1002 – 1007.

Frye v. United States, 54 App. D.C. 46, 293 F. 1013, 1014 (1923).

Frye v. United States, 293 F. 1013, 1014, 34 A.L.R. 145 (D.C. Cir. 1923).

Herscaft, E. E. Manual of Forensic Odontology, Fourth Edition (2011) at page 276.

International Association for Identification, Resolution 97-9 (August 1, 1997).

Mayo, K. “Photo Finish,” Evidence Technology Magazine (July-August 2003, Volume 1, Issue 2).

Rodd v. Raritan Radiologic Associates, P.A., 860 A.2d 1003, N.J. Super. App. Div. (2004).

Staggs, S. “The Admissibility of Digital Photographs in Court,” Crime Scene Investigator Network. Retrieved August 15, 2016 from:

State of California v. Phillip Lee Jackson (1995).

State of Florida v. Victor Reyes, Case No. 99-11535CF10A Memorandum Opinion and Order on Defendant’s Motion in Limine Re: Latent Fingerprint Analysis, signed by the Honorable Stanton S. Kaplan.

The State of Ohio, Appellee, v. HARTMAN, Appellant. Supreme Court of Ohio, Case No. 98-1475, Decided: October 3, 2001. (See more at:

State of Washington, Respondent, v. Eric H. Hayden, Appellant. Court of Appeals of Washington, Division 1, Case NO. 38162-8-I, Decided: February 17, 1998 (See more at:

SWGIT Guidelines, Section 5 Guidelines for Image Processing, SWGIT, Version 2.1 2010.01.15
Warrick, P., “King County Sheriff’s Latent Lab assist in Akron PD homicide investigation,” Pacific NW IAI Examiner (July-December 1999).


< Prev   Next >

Lifting Latent Fingerprints from Difficult Surfaces

ALMOST ANYONE can find, process, and lift a latent print that happens to be in a logical and obvious place like a door handle, a beer can, or a butcher knife. But sometimes, a latent print is not just sitting there in a logical and obvious place. Sometimes, you have to use your imagination to find the print and your skills to lift it.