Transcript Forensic Surface Metrology, Firearms and Tool Mark Evidence

Forensic Surface Metrology
3
2
1
0
1
2
3
Firearms and Tool Mark Evidence
Outline
• Introduction and the Daubert Standard
• Theory of Tool Mark Identification
• Details of Our Approach
• Preliminary results on 9mm cartridge cases
• Nice images of chisel striation patterns
• Future Work
• GPU-CUDA
• Surface alignments/normalizations
• Feature extraction libraries
• Rotation-translation invariants
U.S. v. Brown and U.S. v. Glynn
• Court ruled that ballistics was not a science
• Firearms examiner could not testify to “a
reasonable degree of ballistic certainty”
• Firearms examiner could not “claim conclusions
reached were not to any degree of certainty”
• Firearms examiner could only testify that a
match was “(at least) more likely than not”
• “at least” was ordered dropped in Glynn
Raising Standards with Data and
Statistics
• DNA profiling the most successful application of
statistics in forensic science.
• Responsible for current interest in “raising standards” of
other branches in forensics.
• No protocols for the application of statistics to
physical evidence.
• Our goal: application of objective, numerical
computational pattern comparison to physical
evidence
The Daubert Standard
• Daubert (1993)- Judges are the
“gatekeepers” of scientific evidence.
• Must determine if the science is reliable
• Has empirical testing been done?
• Falsifiability
• Has the science been subject to peer review?
• Are there known error rates?
• Is there general acceptance?
• Federal Government and 26(-ish) States
are “Daubert States”
Forensic Tool Mark Examination
• Most tool usage involves transfer of microscopic
marks onto an impressible surface
Manufacturing Marks
• Marks on tools from:
•
Manufacturing process
• Marks on tools’ working surface continue to change
over time with: Wear and Abuse:
•
Use/wear
•
Abuse
•
Corrosion
Individual, reproducible
tool marks
• Working assumption of the tool mark examiner:
•
Generally tools impart surface features that are unique to
Slide courtesy of Gerard Petillo, Forensic Tool Mark
themselves
Examiner and FBI Firearms/Tool Mark Unit
Tool Mark Comparison Microscope
Current Approach For Striated Tool Marks
• Obtain striation pattern profiles form 3D confocal microscopy
Raw z-heights
3D rendering
Denoised
Primer shear
Glock 19 firing pin impression
• 3D confocal image of entire shear pattern
Shear marks on primer of two
different Glock 19s
Shear mark on different cartridge
casings from same Glock 19
• Surface processing:
• Form removal
• 3rd degree polynomial
• Optional shift skewed
profiles
• Use max CCF
• Filter surface into
waviness and roughness
components
• Cubic spline filter:
s  0.08 mm
Mean total profile:
Mean “waviness”
profile:
Mean “roughness”
profile:
Statistics
s2=4.24
• Treat each profile point like a random variable
• Use Principal Component Analysis to reduce data set
dimensionality
Math in Pictures!!
PC 3
PC 1
PC 2
140D to 3D Screwdriver Striation Patterns
52% variance
Pattern Identification and Error Rates
• Determine efficient decision rules in the absence of any
knowledge of probability densities for the data
• Maximum margins of separation, SVM:
• 3D PCA-SVM Bootstrap error rate ~1%:
Five Consecutively Manufactured Chisels
G. Petillo
Lead impression media
Striation patterns
generated at 32o
70 striation patterns total:
•20 for traditional comparison
•50 for confocal microscopy
G. Petillo
5/8” Consecutively manufactured chisels
Known Match Comparisons
G. Petillo
5/8” Consecutively manufactured chisels
Known NON Match Comparisons
G. Petillo
Just Getting Started: Things to Come
•
•
•
•
•
Dust
Soil
Wrenches
Chisels
Hammers
•
•
•
•
Tire Tracks
Hair
Blood Spatter
Gun Shot
Residue
Future Technical Work
• Offload compute heavy data parallel operations with “massively
parallel” nVidia GPU/CUDA
• Fourier-Mellin for registration
• Noise removal??
• Rotation-Translation Invariant Features
• Hu-invariants, based on central moments:
n
m
~ 8  O nm 
 pq    i  n / 2   j  m / 2  z(i, j)
p
q
i 1 j 1
• Kondor SO(3) invariants pl ,l ,l 
1 2
l
l1
 
ml m1  l1
n1
n2
*
*
2 ,l
Cml11,l,m
f̂
f̂
m1 ,m l1 ,m1 l2 ,m m1 f̂l,m
m
~ O (n1n2 )6  f̂lm    z(i, j) Yl ( ,  )
i 1 j 1
• Others??
Acknowledgements
• National Institute of Justice
• New York City Police Department Crime Lab
• John Jay College of Criminal Justice
• Research Team:
• Helen Chan
• Mr. Peter Diaczuk
• Manny Chaparro
• Ms. Carol Gambino
• Dr. James Hamby
• Aurora Ghita
• Dr. Thomas Kubic
• Eric Gosslin
• Off. Patrick McLaughlin • Frani Kammerman
• Mr. Jerry Petillo
• Brooke Kammrath
• Mr. Nicholas Petraco
• Loretta Kuo
• Dr. Peter A. Pizzola
• Dale Purcel
• Dr. Graham Rankin
• Stephanie Pollut
• Dr. Jacqueline Speir
• Rebecca Smith
• Dr. Peter Shenkin
• Elizabeth Willie
• Mr. Peter Tytell
• Chris Singh
• Melodie Yu