___________________________________
Last week, Iain McKie responds to thought about what happens when a splinter is allowed to fester.

This week We look at a concept introduced a few weeks ago as High Dynamic Range Mapping of images (HDRM).  I sent out a solicitation to the community through the Detail, and found several interested parties.  Although they all expressed interest and the desire to learn more about the subject, one individual stepped up, continued the research and has prepared an excellent Weekly Detail on his findings.  We are also planning on putting together an article for submission to the JFI that has even more detail and photographs than this "pilot", but I think latent print examiners will find this concept extremely interesting and will begin to incorporate this technique into the work flow of latent print examination over the coming years.
___________________________________
High Dynamic Range Fingerprint Images in Photoshop CS2 
by Chad Day and Kasey Wertheim

Introduction

The dynamic range, the ratio between dark and bright regions, exceeds the range of the human eye, printed images, or images displayed on a monitor.  However, one technique to overcome the deficiencies of human biology and technology is by using High Dynamic Range (HDR) images.  A High Dynamic Range image is a fusion of photographs of the same scene, in this case a fingerprint on a surface, taken with different amounts of exposure to light.  By taking multiple photographs of the same image, the photographer is able to record all of the luminance values of the fingerprint.  Normally, an algorithm is employed to recover the film response function and construct the High Dynamic Range (HDR) image.  Fortunately, with the release of Photoshop CS2, the computer now does all the reconstructing.  A fingerprint examiner can easily select which images to merge into an HDR image without dealing with any confusing algorithms.  The program also enables the examiner to freely work with color HDR images.  Although HDR images are currently mostly used in motion pictures and 3D work, HDR images have the potential to be another useful tool for the forensic community.

Photoshop stores the luminance values in a 32-bits-per-channel image, which are directly related to the light in a scene, whereas a 16-bits-per-channel image only stores luminance values from black to white.  Hence, the 32 bit image represents a much larger portion of the actual dynamic range. 

For this pilot study, digital photographs were taken of fingerprints deposited on a glossy black and white book cover treated with gray magnetic powder, a soda can treated with black  magnetic powder, a piece of pink paper treated with ninhydrin, a piece of green paper treated with ninhydrin, and a plastic bag treated with Rhodamine dye stain (viewed under an Argon-Ion laser).  The digital camera used was a Kodak DCS Pro SLR/n connected to the computer via firewire.  The lens was a 60mm Micro Nikon Lens and the ISO of the camera was set at 160 for all photographs.  Initially the images were in RAW format with a resolution of 1000 ppi. 

The items were placed on a copy stand and photographed using ten different shutter speeds using only ambient (fluorescent) lighting.  The aperture and ISO was kept exactly the same for each latent print; varying anything other than the shutter speed can produce lower-quality results.  A properly exposed image was also taken of each item to serve as a comparison to each HDR image.  Except where noted, none of the images were modified using any of the other features of Adobe Photoshop in an effort to examine any potential pre-adjustment improvements gained by using an HDR image over a normally exposed image.  After capture, the digital images were converted to TIFF format and the automatic camera adjustments were removed using the Photoshop RAW file manager. 

After the images were captured and properly adjusted they were merged into an HDR image using the “Merge to HDR” command.  Using the merge command can be done from Photoshop itself by selecting “Choose File”→ “Automate” → “Merge to HDR.”  Click “Browse” in the “Merge to HDR” dialogue box and select the desired images.  Or, the Photoshop Bridge may be used to perform the function by selecting the desired files and clicking “Tools” → “Photoshop” → “Merge to HDR.”  The computer then calculates the camera response curve and presents the user with a preview consisting of thumbnails of the images used in the merge, the 32 bit image itself, a Bit Depth menu, and a slider for setting the white point. 

At this point, Photoshop provides the user with several options.  The Bit Depth menu allows the user to change the image to a 16 or 8 bit image.  However, doing so at this stage would prevent the user from taking advantage of the filters and other tools available in Photoshop while working with the 32 bit image.  Thus, all the images used in the experiment were left as 32 bit images until later in the process. 

It is very important to adjust the white point of the image.  As monitors are generally only capable of handling 24 bits the image will appear distorted when viewing.  By adjusting the white point the user can make the image more viewable.  All test images were adjusted to create the highest quality viewable image using the white point slider.

The last option available in the preview proved to be the most valuable.  The thumbnails of the merged images are presented along the side of the 32 bit image.  By clicking on the boxes below a thumbnail the user can remove that image from being considered in the final HDR image.  Photoshop automatically recalculates the effect on the final image and makes the changes to the preview, allowing the user to see the effect each individual image has on the final image.  This feature was used throughout the experiment to find the best combination of images for each item.  For example, Set B, the images from the soda can treated with black magnetic powder, consists of ten images.  However, HDR B1 is comprised of only four of the ten original images, 142B taken with a shutter speed of 4s, 146B taken at ¼s, 148B taken at 1/15s, and 149B taken at 1/30s. 

Upon finishing with the preview options the computer completes the merge and presents the user with a 32 bit HDR image.  All of the HDR images in the experiment were then converted to 16 bits to make them available for printing.  At the 32 bit to 16 bit (or 8 bit) conversion stage of the process Photoshop provides some useful options in order to preserve as much of the desired dynamic range as possible.  The correction options are “Exposure and Gamma,” “Highlight Compression,” “Equalize Histogram,” and “Local Adaptation.”  “Exposure and Gamma” allows the user to manually adjust the brightness and contrast of the image and was used on HDR A1, D1, E1, and G1.  “Highlight Compression” compresses the highlight values so they fall within the luminance values for the 16 bit image.  “Highlight Compression” did not provide favorable results on any of the test items and was not used during the experiment.  “Equalize Histogram” compresses the dynamic range while attempting to preserve contrast and was used on HDR B1.  “Local Adaptation” adjusts the tonality of the image and was used on HDR C1 and F1.

For a complete set of instructions on using the “Merge to HDR” function of Photoshop CS2 see the Photoshop CS2 User Guide, pages 139-144.

Examples

Example 1
The subject item from Set A was a portion of a glossy, black and white paperback book cover.  The comparison image, Comparison A1, was photographed with an aperture of 4.8 and a shutter speed of 1/15s.  Images 119A and 124A (marked in red below) were ultimately selected to create HDR A1. 


FIGURE 1
The series of images taken for this impression
Highlighted in red are the images used for the HDR image in Figure 1


FIGURE 2
Left: "Before" - A regular photograph of a latent on the item
Right: "After" - An HDR image (A-1) of the same latent on the same item

The ridge detail on the resulting HDR image was much more vivid than that of the comparison image.  In the HDR image, ridge detail could be clearly seen in the black portions of the image.  The same black areas of the comparison image dominate the gray powder.  In addition, the HDR image made the entire print more viewable than the comparison image.

Example 2

The subject item from Set B was a soda can with green, white, and red lettering.  The comparison image, Comparison B1, was photographed with an aperture of 16 and a shutter speed of 1s.  Images 142B, 146B, 148B, and 149B were selected to create HDR B1. 


FIGURE 3
The series of images used to make the HDR image

Again, the resulting image presented a fingerprint with much clearer detail and greater contrast than that obtained by the normally exposed comparison image.


FIGURE 4
Left: "Before" - A regular photograph of a latent on the item
Right: "After" - An HDR image of the same latent on the same item

Set 3

The comparison image, Comparison E1, was photographed with an aperture of 4.5 and a shutter speed of 1/30.  Images 181E, 182E, and 188E were selected to create HDR E1. 


FIGURE 5
The series of images used to make the HDR image

HDR E1 proved to be more vivid and display more contrast than the comparison image.  As an example of how removing one of the images changes the resulting HDR image, image 188E was removed to produce HDR E2.  Deleting image 188E removed most of the background pigment, leaving slightly purple ridge detail on a gray background. 


FIGURE 6
Left: "Before" - A regular photograph of a latent on the item
Right: "After" - An HDR image of the same latent on the same item

Conclusion

HDR images were found to consistently produce detail of higher quality than a normally exposed photograph.  Although taking several photographs at different exposure and merging them together is more time consuming than simply taking a normal digital photo, it could provide the examiner with the ability to see the extra detail needed to identify a tough latent.  Starting with the highest quality image prior to further adjustments in Photoshop gives the examiner more information to work with and enables a higher quality result to be produced.  Hopefully, this study spurs further interest in and use of HDR images by the Latent Fingerprint Community.  Further study and publication on the subject of the use of  HDR image techniques on latent prints is being planned by the author.

References

DEBEVEC, P., MALIK, J.   Recovering High Dynamic Range Radiance Maps from       Photographs.  University of California at Berkeley, Computer Science Division.

ADOBE  CORPORATION   Adobe Photoshop CS2 User Guide, 2005, pp. 139-144.

______________________________________________________________________
Remember, the message board is always open: (http://www.clpex.com/phpBB/viewforum.php?f=2).  For more formal latent print discussions, visit onin.com: (http://www.onin.com)

______________________________________________________________________
UPDATES ON CLPEX.com

No major updates on the website this week.
______________________________________________________________________
Feel free to pass The Detail along to other examiners.  This is a free newsletter FOR latent print examiners, BY latent print examiners. There are no copyrights on The Detail, and the website is open for all to visit.

If you have not yet signed up to receive the Weekly Detail in YOUR e-mail inbox, go ahead and join the list now so you don't miss out!  (To join this free e-mail newsletter, send a blank e-mail to: theweeklydetail-subscribe@topica.email-publisher.com)  Members may unsubscribe at any time.  If you have difficulties with the sign-up process or have been inadvertently removed from the list, e-mail me personally at kaseywertheim@aol.com and I will try to work things out.

Until next Monday morning, don't work too hard or too little.

Have a GREAT week!