NISTIR 8109
Tattoo Recognition Technology – Best Practices
(Tatt-BP)
Guidelines for Tattoo Image Collection
Mei Ngan
George W. Quinn
Patrick Grother
Information Access Division
Information Technology Laboratory
This publication is available free of charge from:
http://dx.doi.org/10.6028/NIST.IR.8109
February 2016
U.S. Department of Commerce
Penny Pritzker, Secretary
National Institute of Standards and Technology
Willie May, Under Secretary of Commerce for Standards and Technology and Director
!
Tatt
Tattoo Recognition Technology - Best Practices
(Tatt-BP)
Guidelines for Tattoo Image Collection
NIST Inter
agency Report 8109
Mei Ngan, George W. Quinn, and Patrick Grother
Image Group
Information Access Division
National Institute of Standards and Technology
February 18, 2016
i
Disclaimer
While this guide provides recommendations on handling acquisition issues commonly seen in operations,
there may exist a small fraction of circumstances that are not covered in this document.
Release Notes
. Typesetting: Virtually all of the tabulated content in this report was produced automatically. This involved
the use of scripting tools to generate directly type-settable L
A
T
E
X content. This improves timeliness, flexibility,
maintainability, and reduces transcription errors.
. Contact: Correspondence regarding this report should be directed to [email protected].
Acknowledgements
The authors would like to thank the sponsor of this activity, the Federal Bureau of Investigation (FBI) Biometric
Center of Excellence (BCOE), for initiating and progressing this work.
The authors are also grateful to the following reviewers who provided valuable input to this document:
- Amanda Noxon (Michigan State Police)
- Brad Wing (NIST)
- Daniel Manger (Fraunhofer IOSB)
- Darin Walker (FBI)
- Edward Delp (Purdue University)
- Eric Phillips (FBI)
- Jacob Ruberto (Pinellas County Sheriff’s Office)
- Landis Huffman (MITRE)
- Lisa Goff (FBI)
- Mark Gaus (FBI)
- Mike Garris (NIST)
- Nick Orlans (MITRE)
- Peter Lo (Morpho)
- Steve Tomanovich (Compass Technical Consulting)
- Steven Carter (FBI)
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
CONTENTS ii
Contents
DISCLAIMER I
RELEASE NOTES I
ACKNOWLEDGEMENTS I
1INTRODUCTION 1
1.1 AUDIENCE ............................................................ 1
2THE TATTOO RECOGNITION TECHNOLOGY PROGRAM 1
3NUMBER OF PHOTOGRAPHS 2
3.1 SINGLE TATTOOS ........................................................ 2
3.2 TATTOOS ON MULTIPLE BODY LOCATIONS .......................................... 2
3.2.1 DISTINCTLY SEPARATE TATTOOS ............................................ 2
3.2.2 MULTIPART TATTOOS .................................................. 3
3.2.3 FULL-BODY TATTOOS .................................................. 3
4IMAGE SIZE 4
5STANDOFF DISTANCE 4
6COMMON COLLECTION PROBLEMS 4
6.1 BACKGROUND .......................................................... 4
6.2 ILLUMINATION ......................................................... 4
6.3 CLOTHING ............................................................ 5
6.4 ORIENTATION/BODY POSI TIONING .............................................. 5
6.5 FOCUS .............................................................. 6
6.6 SENSOR PLANE ......................................................... 6
6.7 BODY HAIR ........................................................... 7
7TRANSMISSION 7
8SUMMARY OF RECOMMENDATIONS 8
List of Figures
1PROGRAM OVERVIEW ...................................................... 1
2NUMBER OF PHOTOS ...................................................... 2
3MULTIPLE TATTOOS ....................................................... 3
4MULTI-PART TATTOOS ...................................................... 3
5FULL-BODY TATTOOS ...................................................... 3
6BACKGROUND .......................................................... 4
7ILLUMINATION ......................................................... 5
8CLOTHING OCCLUSION ..................................................... 5
9CLOTHING FIELD OF VIEW ................................................... 5
10 ORIENTATION .......................................................... 6
11 FOCUS .............................................................. 6
12 SENSOR PLANE ......................................................... 6
13 BODY HAIR ........................................................... 7
14 GOOD QUALITY TATTOO IMAGE ................................................. 8
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
1
Tattoo Recognition Technology (Ta t t ) Program
Ta t t -C
“Open-book” challenge with
public dataset to engage research
community to advance image-
based tattoo matching technology
Ta t t -BP
Best practice guidelines and
training material for the collection
of tattoo images
Ta t t -E
Large-scale sequestered
evaluation of tattoo recognition
algorithms
2014 - 2015
2015 - 2016
2016 - 2017
“Open-book” test with public
dataset
Best Practice Documents Sequestered Test
Future Activities
TBD
TBD
Comments and
discussion welcome via
Figure 1: Activities under the Tattoo Recognition Technology Program, including planned future projects. See website http://www.
nist.gov/itl/iad/ig/tattoo.cfm for latest project status.
1 Introduction
This document, along with a supplementary poster [3] and a set of instructional slides [2], provides best practice guidelines
for the collection of good quality tattoo images. As an outcome of the Tatt-C 2015 [4] activity, algorithm failure to correctly
match a tattoo is often related to the consistency and quality of image capture. Notably, inconsistencies in image angle,
orientation, size of the tattoo relative to the entire image and poor collection characteristics such as poor illumination,
low contrast, blur, and the existence of clothing and background clutter caused failures for tattoo detection and matching
algorithms. While some problems can potentially be rectified via post-capture image processing, which imposes additional
human labor, certain properties cannot be recovered after the photograph is taken. As such, certain guidelines should
be followed to ensure good quality images are collected. For the purpose of this document, for specific photographic
guidance that is not detailed in this document, the recommendations of ANSI/NIST-ITL Standard [1] Annex E (Facial
Capture) apply.
1.1 Audience
The intended audience of this document is law enforcement professionals or officers who photograph tattoos and/or
design or specify image collection processes. The recommendations provided are simple and straightforward. Although
the number of ways image collection can go wrong might appear quite numerous, a single adjustment can often rectify
several problems at once. Thus, the operator only needs to remember a few simple guidelines to deal with the majority of
problems that might occur.
2 The Tattoo Recognition Technology Program
The Tattoo Recognition Technology Program (Figure 1) was initiated by NIST to support an operational need for image-
based tattoo recognition and classification to support law enforcement and forensic applications. The program provides
quantitative support for tattoo recognition development and best practice guidelines. A summary of past, ongoing, and
planned activities is provided below.
• Tatt-C [4] was an initial research challenge that provided operational data and use cases to engage the research com-
munity into advancing research and development into automated image-based tattoo technologies. It also sought
to assess the state-of-the-art to determine what methods are effective and viable for pertinent operational scenarios.
NIST hosted a culminating industry workshop and published a public report on the outcomes and recommenda-
tions from the Tatt-C activity. While the Tatt-C participation period has closed, the Tatt-C dataset is available to
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
2
developers on an ongoing basis to support research and development. Please visit http://www.nist.gov/itl/
iad/ig/tatt-c.cfm for more information.
• Tatt-BP provides best practice guidance material (including this document) for the proper collection of tattoo images
to support image-based tattoo recognition. Recognition failure in Tatt-C was often related to the consistency and
quality of image capture, and Tatt-BP aims to provide guidelines on improving the quality of tattoo images collected
operationally.
• Tatt-E will be a sequestered evaluation intended to assess tattoo recognition algorithm performance over a large-
scale operational dataset.
3 Number of Photographs
3.1 Single Tattoos
A minimum of two photographs should be taken of a tattoo - one photo that is far enough away to expose the body
location of tattoo and one close-up photo in which the tattoo occupies at least 75% of the image (i.e., tattoo-to-image ratio
of 75%) (Figure 2). Capturing body location context in a tattoo image has forensic and investigative value while a zoomed
close-up of the tattoo may aid automated tattoo recognition software. In the event that a close-up photograph is not
collected, manual cropping of the image (e.g., using an image editor) is an acceptable alternative. There is a field in the
ANSI/NIST-ITL [1] Type-10 record (Field 10.039 Type-10 reference number) used to link two or more images of the same
tattoo by using the same index reference for all of the associated images in a transaction.
#1! #2!
#1!
#2!
#1! #2!
Figure 2: Examples of photos far enough away to expose body location (#1) and close-up photos (#2).
3.2 Tattoos on Multiple Body Locations
3.2.1 Distinctly Separate Tattoos
For distinctly separate tattoos on different body locations, separate photographs for each tattoo should be taken, which
includes a far away photo exposing the body location and a zoomed close-up (Figure 3).
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
3.2 TATTOOS ON MULTIPLE BODY LOCATIONS 3
Figure 3: Examples of collecting images for multiple tattoos on distinctly separate body locations.
3.2.2 Multipart Tattoos
For tattoos that span multiple body locations, and it is not possible to capture the entire tattoo from a single angle, collect
multiple images from different location viewpoints, with overlap where possible (Figure 4).
Front!
Side!
Back!
Figure 4: Example of collecting a tattoo that spans multiple body locations from different location viewpoints.
3.2.3 Full-body Tattoos
For large or full-body tattoos, in addition to collecting an image of the large tattoo, collect images of smaller portions of the
tattoo (such as a detail of the tattoo that could be useful for determining gang affiliation). Each significant tattoo portion
should be imaged separately, along with an image of the entire tattooed area (Figure 5).
#1!
#2!
#3!
Figure 5: Example of collecting a full-body tattoo, with one photo that captures the entire body (#1) and multiple close-up photos of
areas of interest (#2 and #3).
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
4
4 Image Size
Collect tattoo images at least 1920 x 2560 (or 2560 x 1920) pixels or larger. This is equivalent to a photo resolution of
5 megapixels. Per the ANSI/NIST Standard [1] Annex E recommendations for facial capture compression algorithm,
images shall be collected in JPEG format with a maximum compression ratio of 15:1 or less. The high resolution supports
potential post-collection zooming or cropping of a wide range of tattoo-to-image ratios.
5 Standoff Distance
The subject should be at a minimum of one meter (approximately 3.3 feet) or more away from the camera to mitigate
perspective distortion. The zoom lens settings on the camera should be adjusted accordingly to achieve the desired tattoo-
to-image ratio for the close-up photo.
6 Common Collection Problems
This section describes the most commonly encountered collection problems that lead to poor quality tattoo images. Rec-
ommendations on identifying and rectifying each are provided.
6.1 Background
The discussion of background is important for tattoo recognition, because the first step in the tattoo recognition process
is segmentation of the tattoo from the background. Where possible, the image should only contain the tattoo and the
surrounding skin and should not contain anything else in the background, including patterned walls and floors, furniture,
clothing, or other body parts. If elimination of the background is not possible, the background should be a plain, solid
color in order to prevent pattern distractions that could interfere with tattoo segmentation (Figure 6).
Figure 6: Examples of incorrect and correct background for photo collection.
6.2 Illumination
A tattoo image should be taken under uniform, diffuse lighting. Diffuse light is non-directional light, where the intensity
of light is evenly distributed. Diffuse lighting produces very uniform, soft lighting which illuminates the entire space and
makes objects visible, but avoids large shadows and hot spots/reflections (Figure 7). It is recommended that the room
light level be at least 500 lux (e.g., bright office lighting). Per the ANSI/NIST-ITL Standard [1] Annex E recommendations
for facial capture lighting, subject illumination should be accomplished using a minimum of three point-balanced illumi-
nation sources. Although a minimum of three photo lights is required, two of these lights should be sufficient for some
operational environments. In cases where there is low contrast between the tattoo and the skin (e.g., black ink on dark
skin or white ink on light skin), consider using an additional source of lighting to bring out the contrast between the tattoo
and the skin. Typically, camera flash should not be used due to large light reflections that may be introduced.
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
6.3 CLOTHING 5
Figure 7: Example photos with non-uniform or poor illumination versus exemplary photos of tattoos with adequate, uniform lighting.
6.3 Clothing
The entire tattoo should be in view without any parts being covered by clothing (e.g., t-shirt sleeves, wristbands, etc.).
Clothing, especially patterned clothing, should be eliminated from the camera’s field of view prior to image capture.
When possible, move obstructive clothing so the entire tattoo is in view (Figure 8). If clothing elimination is not possible,
adjust camera position and zoom settings to eliminate clothing from the camera’s field of view (Figure 9).
Figure 8: Example photos of clothing covering up parts of the tattoo (problematic) versus exemplary photos of tattoos with no clothing
in view (ideal).
Figure 9: Examples of the existence of clothing in far away photo but adjusting camera zoom to eliminate clothing from the camera’s
field of view in zoomed close-up.
6.4 Orientation/Body Positioning
To ensure orientation consistency, tattoos on body limbs should be captured with the body part parallel to the torso. For
example, a tattoo on a forearm should be photographed with the arm pointing towards the ground. A tattoo on a leg
should be collected with the subject standing upright (Figure 10).
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
6.5 FOCUS 6
Figure 10: Examples of incorrect versus correct body orientation.
6.5 Focus
The entire tattoo should be in clear focus, and no part of the tattoo should be blurry (Figure 11). Motion blur occurs when
the camera or the subject is moving and can be mitigated or rectified by holding the camera still and ensuring the subject
remains still during photo capture. Focus blur usually occurs, because the subject is too close or too far from the camera.
This can be mitigated or rectified by ensuring the subject is at the proper distance from the camera and by adjusting the
camera lens focus.
Figure 11: Examples of out-of-focus (bad) versus in-focus (good) photos.
6.6 Sensor Plane
Tattoos should always be captured parallel to the camera sensor plane. This means the tattoo should be in straight view
and not at any vertical or horizontal angle. The camera should be pointing straight at the tattoo and not from a vertical
or horizontal angle (Figure 12). If the tattoo is too large and parts of the tattoo aren’t parallel to the camera sensor plane
(often on arms and legs), then collect from multiple angles according to Section 3.2.2.
Figure 12: Examples of incorrect versus correct sensor plane alignment.
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
6.7 BODY HAIR 7
6.7 Body Hair
Hairy regions that are not a part of the tattooed region should be avoided as hair can introduce noise in the image that can
affect tattoo recognition. If irrelevant body hair cannot be avoided in the photograph that exposes the body location, then
ensure that body hair is eliminated from the camera’s field of view in the close-up image (Figure 13).
Figure 13: Examples of hairy regions near tattooed regions (outlined in red) and eliminating body hair from camera’s field of view in
zoomed close-up photograph.
7 Transmission
Post-collection, we recommend storage and transmission of tattoo images to law enforcement agencies (e.g., FBI) using the
ANSI/NIST-ITL Standard [1] Type-10 record. The Type-10 record contains additional supporting metadata fields such as
body location and tattoo descriptors and is available at http://www.nist.gov/itl/iad/ig/ansi_standard.cfm.
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
8
8 Summary of Recommendations
This section provides a brief summary of the recommendations presented earlier on how to collect good quality tattoo
images.
• Capture Environment: Ensure there is adequate, uniform, diffuse lighting (ideally with a minimum of two point-
balanced light sources and lighting level of 500 lux). The background should be a plain, solid color and does not
contain any background clutter.
• Prior to Capture: Verify the camera is set to collect images at a resolution of 5 megapixels or higher. Ensure there is
a minimum of one meter (approximately 3.3 feet) between the subject and the camera. Instruct the subject (if able) to
stand upright with both forearms pointing towards the ground and remain still. Ensure the entire tattoo is in view
without any parts being covered by clothing.
• During Capture: Ensure the camera is held still and properly aligned to capture the tattoo parallel to the sensor
plane. For each tattoo, collect a minimum of two photographs - one that is far enough away to expose the body
location and one close-up photo where the tattoo occupies at least 75% of the image and ideally does not include
any clothing, body hair, or background clutter. For tattoos that span multiple body locations, collect multiple images
from different location viewpoints with overlap where possible. For full-body tattoos, collect an image of the entire
tattoo, then separate photos of smaller areas of interest.
• Post-Capture: Ensure the tattoo is in focus, with good illumination and contrast against the skin. Ensure there are
no large shadows or reflections over the tattoo.
Figure 14 shows an example of a good quality frontal image of a tattoo.
Plain, solid!
background!
In focus, good !
contrast against
the skin !
Adequate, uniform
lighting!
No clothing or body
hair in view!
No shadows
or reflections
over tattoo!
Tattoo is parallel to
camera’s sensor plane!
Figure 14: Example of a good quality frontal image of a tattoo, with some of its desirable properties marked up.
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
REFERENCES 9
References
[1] ANSI/NIST-ITL 1-2011:Update 2015, NIST Special Publication 500-290, Data Format for the Interchange of Fingerprint,
Facial and Other Biometric Information. http://www.nist.gov/itl/iad/ig/ansi_standard.cfm.
[2] NIST Tattoo Recognition Technology - Best Practices (Tatt-BP) Guidelines for Tattoo Image Collection (Instructional
Slides). http://biometrics.nist.gov/cs_links/tattoo/tatt-bp/Tatt-BP_Instructional_Slides.
pdf.
[3] NIST Tattoo Recognition Technology - Best Practices (Tatt-BP) Guidelines for Tattoo Image Collection (Poster). http:
//biometrics.nist.gov/cs_links/tattoo/tatt-bp/Tatt-BP_Poster.pdf.
[4] M. Ngan, G. W. Quinn, and P. Grother. NISTIR 8078 Tattoo Recognition Technology Challenge (Tatt-C) Outcomes
and Recommendations., September 2015. http://www.nist.gov/itl/iad/ig/tatt-c.cfm.
TATT-BP: GUIDELINES FOR TATTOO IMAGE COLLECTION
