Facial Muscle Anatomy Based Approach for Forensic Facial

Transcription

Facial Muscle Anatomy Based Approach for Forensic Facial
Facial Muscle Anatomy-based Solution
for Forensic Facial Reconstruction in
Sri Lanka
University of Colombo School of Computing (UCSC)
Research Team
Group Members :
Supervisors
:
Anuradha.K Madugalla
Roshan. N Rajapakse
Ishari .U Amarasinghe
Vinavi .H Padmathilake
Dr. Anuja Dharmaratne
Mr. K D Sandaruwan
Dr. M.Vidanapathirana
Mass Disasters
Forensic Identification
End of Process
Problem Statement
Facial Reconstruction is not technically implemented in SriLanka due to :
 Lack of Facial Tissue Thickness Data
 Lack of Facial Feature Data
 High Cost of Purchasable Solutions
Even the Manual Method is not implemented due to :
 Requires a lot of time
 Less productive
 Lack of Expertise
The Objectives
To introduce a feasible
Computer based
three dimensional
facial reconstruction solution to Sri Lanka
To Conduct
Tissue Thickness Analysis
Facial Component Analysis; for the First time in Sri Lanka
Facial Reconstruction Types
Facial
Reconsturction
3D Reconstruction
2D Reconstruction
Manual
Sketch
Computer
assisted
Sketch
3D computer
graphics based
Modeling
Manual
Methods
Anthropometrical
Anotomical
Combination
Sparse
Approach
Dense
Approach
The Manual Method
1. Anthropometrical
2. Anatomical
Abate, A. et al. (2004) FACES: 3D Facial reConstruction from anciEnt Skulls using
content based image retrieval. Journal of Visual Languages and Computing, vol. 15, pp. 373-389.
3D computer graphics based Modeling
Andersson B. and Valfridsson M.(2005) Digital 3D Facial Reconstruction Based on Computed Tomography, Master’s Thesis, Linkopings University
Davy et al. (2005) Computer-Graphic Facial Reconstruction:Forensic facial reconstruction using computer modelling software. Elsevier, pp.183194.
Semi-Automation
Reconstruction Method
Advantages
Disadvantages
Manual
Reconstruction
 Natural looking
reconstructed face
 Manual Skull models
 Clay sculpting
Automated
Reconstruction
 Quick skull models
 Clay material not
needed
 Unrealistic
Reconstructed Face
Our SolutionSemi Automated
Reconstruction
 Natural looking
reconstructed face
 Quick skull models
 Clay material not
needed
Solution Design
Pre-requisite
Step 01
Step 02
Step 03
Step 04
: Forensic Analysis
: Acquiring the 3D model
: Placement of Landmarks
: Digitally Sculpting the Face
: Adding Facial Components
Pre Requirement: Skull Analysis
Determine
 Sex
 Age
Performed by
Forensic Analysts
Step 01:Skull Input
 3D Scanner(Konica Minolta VIVID 910)
 CT Scan (DICOM)
Step 02:Skin Depth Markers
Face Building
Studying Facial Muscle Anatomy
Step 03: Procedure of Face Creation
Step 04: Adding different Facial components/
Features
Most common Facial Features- Facial Component Analysis
The Process
3D Editing Software
Digital Sculpting Software
Analysis performed Locally
 Tissue Thickness Analysis –Length of Landmarks
 Facial Feature Analysis - Most common Facial Features of Sri
Lankans
No data on Sri Lankans
Tissue Thickness Data Gathering Procedure
 Department of Radiology at
 Colombo South Teaching Hospital
 Central Hospital (Pvt) Ltd
 Durdans Hospital (Pvt) Ltd
 Age range: From 20 to 30
 Weight: Medium weight
Tissue Thickness Measuring
• Different studies-different number of markers
Facial Tissue Thickness Analysis: Observations
 Considerable difference between
different races
 Sri Lankans have greater tissue
thicknesses at points 10, 19, 20 and
21
 Hence, New Facial Tissue Thickness
data gathering is needed- Sri Lankan
Based
Facial Component Analysis
 Determine the most frequent indexes within the particular age
and gender
 Conducted the survey
 Age: 20 – 30 years
 Sample: 500 male & female photographs
 Components: Eyes and nose
The Process
Capture
Skull
Add Markers
Tissue Thickness
Analysis
Muscle
Sculpting
Facial Feature
Analysis
Final Model
Case 01
Actual Photograph of the Deceased
Reconstructed 3D Face Model
Case 02
Actual Photograph of the Deceased
Reconstructed 3D Face Model
Case 03
CT rendered photograph of the patient
Reconstructed 3D Face Model
Case 04
CT rendered photograph of the patient
Reconstructed 3D Face Model
Qualitative
Methods
Face Pool
Comparison
Resemblance
Rating
Assessment
Evaluation
Methods
Quantitative
Methods
Photogrammetry
Analysis
Superimposition
Stephan CN and Henneberg M. Building faces from dry skulls: are they recognized above chance rates? J Forensic Sci 2001;46(3):432–440.
Qualitative Methods : Face Pool Comparison
Qualitative Methods : Face Pool Comparison
Position of the Photograph
Results of Case 1
5th Person
4th Person
3rd Person
2nd Person
1st Person
0
20
40
60
80
100
Number of Votes %
http://facereconsurvey.appspot.com/
Qualitative Methods : Face Pool Comparison
Position of the Photograph
Results of Case 2
4th Person
3rd Person
2nd Person
1st Person
0
20
40
60
80
100
Number of Votes %
http://facereconsurvey2.appspot.com/
Qualitative
Methods
Face Pool
Comparison
Resemblance
Rating
Assessment
Evaluation
Methods
Quantitative
Methods
Photogrammetry
Analysis
Superimposition
Quantitative Methods : Resemblance Rating Assessment
1
• Not at all Similar
2
• Not Similar
3
• Identifiable
4
• Fairly Similar
5
• Very Similar
Quantitative Methods : Resemblance Rating Assessment
Results of Case 3
4.5
Degree of Resemblance
4
3.5
3
2.5
2
1.5
1
0.5
0
Overall Face
The Nose
The Mouth
General Shape
of the face
http://facereconsurvey3.appspot.com/
Quantitative Methods : Resemblance Rating Assessment
Results of Case 4
4.5
Degree of Resemblance
4
3.5
3
2.5
2
1.5
1
0.5
0
Overall Face
Nose
Mouth
General Shape of
the face
http://facereconsurvey4.appspot.com/
Qualitative
Methods
Face Pool
Comparison
Resemblance
Rating
Assessment
Evaluation
Methods
Quantitative
Methods
Photogrammetry
Analysis
Superimposition
Quantitative Methods :Photogrammetry Analysis
Landmark
Abbreviation
Midline landmarks
1
Nasion
N
2
Subnasale
Sn
3
Labrale superius
Ls
4
Menton
Me
Bilateral landmarks
5, 6
Endocanthion
En
7, 8
Exocanthion
Ex
9, 10
Alare
Ala
11, 12
Cheilion
Ch
Quantitative Methods :Photogrammetry Analysis
Results of Case 1
Quantitative Methods :Photogrammetry Analysis
Results of Case 3
Qualitative
Methods
Face Pool
Comparison
Resemblance
Rating
Assessment
Evaluation
Methods
Quantitative
Methods
Photogrammetry
Analysis
Superimposition
Quantitative Methods : Superimposition
Results of Case 1
Quantitative Methods : Superimposition
Results of Case 3
Uniqueness of Our Solution
 Introducing Facial Reconstruction to Sri Lanka
 A Novel Facial Muscle Sculpting based reconstruction
method
 Optimized Marker Placements
 A Solution unique to Sri Lanka - Sri Lankan data (Tissue
Thickness, Facial Feature) based solution
Objectives Achieved
Problem 01 :
– Lack of Tissue thickness data on Sri Lankans
• Solution
 Formulated a Facial Tissue Thickness model for Sri Lankans
Problem 02 :
– Lack of Facial Feature data on Srilankans
• Solution
 Carried out a Facial Feature Analysis on Sri Lankans
Objectives Achieved (Contd..)
•
Problem 03 :
– Expert knowledge on facial reconstruction process
• Solution
 3D Sculpting based simplified process
•
Problem 04 :
– Time Consuming (2-3 weeks)
• Solution
 Less time is required (4-5 hours)
Difficulties Faced
 Difficulty in acquiring a photograph of the deceased for evaluation
purposes
 Non availability of tissue thickness data
Suggestions for Future Research
 To conduct facial tissue thickness analysis within all the age groups in
the Sri Lankan context
 To carry out a facial component analysis among Sri Lankan’s including
all the age ranges
 To integrate the facial reconstruction methodology with the missing
persons database
 Archaeological studies
The way forward….
 Separate tissue thickness research by a team of
consultants
 More digital sculptors
First Forensic Facial Reconstruction Unit
for Sri Lanka
Concluding Remarks
Relatives and
friends of the
Missing
People
Archeologists
Police
Investigators
Forensic
Medical
Officers
Thank You !!!