A Dynamic Range for Impression Evidence
Written by Adam Sabolich   

Adobe Photoshop has become a ubiquitous part of the law enforcement digital imaging community, and has cemented itself as an essential forensic tool for impression evidence examiners and crime scene technicians. The focus of this article is to demonstrate Adobe Photoshop’s high dynamic range function (Merge to HDR Pro), which uses multi-exposure images to enhance and visualize indented writing. While this paper focuses on questioned document examination, the use of merging multi-exposure images is applicable for all forensic sciences involving impression evidence, especially hard-to-photograph items, such as questioned documents, latent prints, footwear, overall crime scene photographs, and more.

This article describes the concept and techniques of digitally processing multi-exposure crime scene images. To validate this technique, 50 images were photographed of an indented writing sample and its subsequent pages. Adobe Photoshop CC 2017 was used to assess varying angles of light sources along with several digital image processing techniques to determine if the visualization of the writing in the photographs could be improved. The most ineffective of the results consisted of two light sources from opposite sides that tended to wash out the detail. The most effective results included digitally merging light sources from the top left and bottom left, as well as highlighting shadows and highlights in blue and red to enhance specific words and letters within the image. While none of the techniques offer a complete “fix-all” solution, certain techniques did subjectively enhance portions of the image.

Many techniques are used currently by forensic scientists to photograph impression evidence. Typically, when photographing indented writing, a light source is held in a low oblique position diagonal to the paper. A similar technique is often employed with footprints, except rather than held diagonal to the impression, a flash is held perpendicular to the footprint. The purpose of these techniques is to accentuate the shadows of the indentations so that they may be properly visualized in a photograph.

While using these concepts in practice, a number of questions occurred to me: After I photograph this indented writing, what if the writing is still illegible? What if part of the writing can be visualized, but certain words or letters cannot? If I used a light source from more than one angle while photographing the subject, would the writing be easier to read? If I can do this while photographing the subject, could I instead digitally overlay separate images of the same subject but with the light source originating in different corners? By describing this process, I hope you will better understand the purpose and goal behind this experiment.

The first step was creating an indented writing sample. With a ballpoint pen, I wrote the following to simulate a note that might be left at the scene of a bank or store robbery:

“STAY CALM. I WANT ALL LOOSE $100 + $20 BILLS FROM THE REGISTER.
NO POLICE, DON’T PANIC. THANKS.”


Figure 1—Front of page 1 with original writing

All photographs were captured on a tripod in an enclosed room with no windows or ambient light, and every photograph was taken with an aperture of f/11, a focal length of 32mm, and an ISO (International Standards Organization) setting of 100. The front and back of the subsequent four pages were photographed as well as the top page with the original writing. A separate photograph was taken with the light source originating in each corner, the exposure of each of these images bracketed at -1, 0, and +1. Therefore, each side of each paper was photographed a total of 12 times.

As the Merge to HDR Pro function in Photoshop was used to process nearly every image in this experiment, I would like to discuss some of the specifics of Merge to HDR Pro, how it was used in this experiment, and what it can offer a digital image processing technician. The “Merge to HDR Pro” function is accessed by selecting Automate from the File menu, and then selecting Merge to HDR Pro. A dialogue box is displayed prompting the user to select the images to be merged. These can be browsed from the hard drive or the option Add Open Files may be used to merge files already open in Photoshop. Upon clicking OK, the Merge to HDR Pro dialogue box is displayed. While there are many settings that may be adjusted in this window, do not adjust exposure, recovery, fill light, blacks, or other settings before entering Merge to HDR Pro because these settings will be zeroed out so that all source images have the same response curve.

One of the advantages of HDR imaging is that an HDR file can be as large as 32 bits, giving the user lots of information with which to work. By default, the Merge to HDR Pro dialogue box will downsample the image to 16 bits. Because the dynamic range of a 32-bit HDR image is so wide, a standard computer monitor cannot display the entire histogram within Photoshop at once without scrolling to different portions. If the user wishes the image to remain at 32 bits, this option can be selected in the Mode dropdown menu. The conversion method for these images was set to 16-bit and Local Adaptation because all slider adjustments for tone mapping are available within this mode. Because images are converted from 32-bit to 16-bit, it is crucial to ensure that the original images are saved and that only copies of the images are used for this process.

Presets within the Merge to HDR Pro dialogue box include three kinds of presets: monochromatic, photorealistic for expanding dynamic range, and surrealistic. Adjusting the Shadow slider to the left increases shadows, moving the Highlights slider to the right increases highlights, and adjusting the Gamma slider to the left increases contrast. The Exposure slider brightens the entire image when moved to the right. If the goal is to increase color intensity, the Saturation slider can be used, although it clips colors; therefore, the Vibrance slider is preferred. Additionally, Curve can be used to further adjust contrast by manually creating and moving points on the histogram. There is also a section marked Edge Glow, where the Radius specifies the size of the local brightness regions, and Strength specifies how far apart two pixels’ tonal values must be before they are no longer part of the same brightness region. There is also a checkbox labeled Edge Smoothness, which smooths out these effects. Subjectively, Edge Smoothness seemed to smooth out the background paper details that could interfere with an attempt to read the indented writing, so this option was typically selected (Sabolich, 2016).

Using the Merge to HDR Pro function, several photograph combinations were created using multiple exposures and varying origins of the light source. Because I merged and digitally processed many of these images, I would like to focus on the least effective and most effective. A merged image (Figure 4) was created of the front side of the second page by combining two images (Figures 2 and 3) with the light source originating from opposite sides. A more effectively visualized merged image (Figure 6) was created by overlaying two images (Figures 2 and 5) with the light source in the first image originating from the bottom left, and in the second image from the top left. This more legible image was also inverted (Figure 7) to assess if this would aid in the visualization of the writing. Finally, a duplicate was made of the image shown in Figure 7, and the image was inverted. I used the Hue/Saturation function on the inverted image to colorize the image as blue, and the noninverted image was colorized to red. As shown in Figure 8, these two colorized images were then overlaid. This process was also completed with the inverted image colorized to red, and the noninverted image colorized to blue before both images were merged (Figure 9).


Figure 2—(Above Left) Front of page 2 with light source originating from bottom left; Figure 3—(Above Right) Front of page 2 with light source originating from top right; Figure 4—(Bottom) Merged image combining Figure 2 and Figure 3

 


Figure 2—(Above) Front of page 2 with light source originating from bottom left; Figure 5—(Above) Front of page 2 with light source originating from top left; Figure 6—(Left) Merged image combining Figure 2 and Figure 5


Figure 7—Inverted version of Figure 6


Figure 8—Merged image with shadows colorized red and shadows colorized blue


Figure 9—Merged image with shadows colorized blue and shadows colorized red

The least effective technique was merging images with the light sources in opposite corners (Figure 4). The opposing sources of light essentially washed out one another, making the visualization worse than the preprocessed images. Using light sources from the top left and bottom left were most effective when the images were merged (Figure 6). This emphasized additional shadows and highlights without washing each other out in the process. The merged image appears even more legible when inverted (Figure 7). As shown in Figure 8, the disparity between the shadows and highlights was further enhanced by colorizing the shadows to red and the highlights to blue, as well as the shadows to blue and the highlights to red (Figure 9).

While visualization of any digital image may be considered subjective, some areas of the image in question became easier to visualize using a combination of these techniques. Using blue for the shadows and red for the highlights appeared to make the overall image more legible, although certain letters and words were more easily visualized using the opposite color scheme. An additional step in processing the colorized images was shifting the foreground image a few pixels horizontally and vertically, thus creating a more three-dimensional effect to visualize the indentations more effectively.

In conclusion, I would also like to point out that while I found these techniques to be quite effective, additional testing should be completed to develop best practices when using these techniques for hard-to-photograph objects. The concept of digitally processing photographs of impression evidence is largely unexplored, and it is my hope that the processes described in this article can be reproduced and improved upon by others to aid in the improved visualization of indented writing as well as other types of impression evidence.

Reference

Sabolich, A.R. Applications of High Dynamic Range Digital Image Processing to Indented Writing, the George Washington University, Washington, D.C. (December 17, 2016).

About the Author

Adam Sabolich

 

 
< Prev   Next >






New Books

Bloodstain Pattern Analysis

Most forensic disciplines attempt to determine the “who” of a crime. But bloodstain pattern analysis focuses on the “what happened” part of a crime. This book is the third edition of Blood-stain Pattern Analysis. The authors explore the topic in depth, explaining what it is, how it is used, and the practical methodologies that are employed to achieve defensible results. It offers practical, common-sense advice and tips for both novices and professionals. www.crcpress.com

Read more...