Get documents about "ANPR Surv
Shared by: HC120911012336
-
Stats
- views:
- 11
- posted:
- 9/10/2012
- language:
- Unknown
- pages:
- 32
Document Sample
The Covert Implementation
of Mass Vehicle Surveillance in Australia
Roger Clarke
Xamax Consultancy, Canberra
Visiting Professor at ANU, UNSW, and the Uni. of Hong Kong
Chair, Australian Privacy Foundation
http://www.anu.edu.au/Roger.Clarke/....
..../DV/ANPR-Surv {.html,.ppt}
Social Implications of Covert Policing Workshop – 7 April 2009
Copyright
1
2008-09
Red Light and Speed Cameras
Quic kT i me™ and a
T IFF (Unc ompres s ed) dec ompres sor Quic kTime™ and a
are needed to s ee this pic ture. TIFF (Uncompres sed) dec ompress or
are needed to s ee this pic ture.
Copyright http://en.wikipedia.org/wiki/Speed_cameras_in_Australia 2
2008-09
Cynicism about Red Light and Speed
Cameras
QuickTime™ and a
TIFF (Uncompressed) decompre ssor
are neede d to see this picture.
Qu ic kTime ™ and a
TIFF (Unco mpres se d) d ec ompres so r
are n eed ed to se e thi s pi cture.
QuickTime™ and a
TIFF (Un compressed) decompressor
are neede d to see this picture.
QuickTime™ and a
TIFF (Uncompressed) decompre ssor
are neede d to see this picture.
Copyright http://www.speedcam.co.uk/ 3
2008-09
http://fightfines.info/ (Vic)
Covert Implementation of Mass Vehicle Surveillance
AGENDA
• Red-Light / Speed Cameras to ANPR
• Traffic Applications
• Blacklist-in-Camera Architecture
• Quality Factors
• LEAs – Operational Applications
• LEAs – Intelligence Applications
• Mass Surveillance ANPR
• ANPR Deployments in Australia
• ANPR Coordination in Australia
Copyright
4
2008-09
Beyond Red Light / Speed Cameras
To Vehicle Surveillance
• Vehicles can be monitored in various ways, e.g.
• Manual Inspection of VINs, registration plates
• Passive RFID-tags passing control-points
• On-Board Transmitters, with self-reporting
of GPS-based or other coordinates
• Vehicle Registration Data can be monitored:
• Cameras were wet chemistry, are now digital
• Data Extraction was manual, is now automated
• Auto-Lookup of Blacklists is now feasible
Copyright
5
2008-09
Automated Number Plate Recognition (ANPR)
QuickTime™ and a
TIFF (LZW) decompressor
are neede d to see this picture.
Copyright
6
2008-09
Automated Number Plate Recognition (ANPR)
• A Digital Camera
Captures an image of a motor vehicle’s 'number' plate
• Software
Extracts the registration data (numbers, letters, perhaps
other data such as colour and jurisdiction identifiers)
• List(s) of Numbers Being Sought
Enables evaluation of the significance of the extracted data
• Transmission Facilities
Sends the extracted data and perhaps other data elsewhere
Copyright
7
2008-09
Traffic Applications
√ Traffic Law Enforcement. Detection and prosecution for:
√ running red lights
√ driving at a point-in-time speed over the speed limit
√ Traffic Law Enforcement. Detection and interception of:
√ Unregistered Vehicles
? Driving at an average speed over the speed limit
?? Vehicles owned by currently Unlicensed Drivers
√ Public Safety. Deterrence of unsafe practices, e.g.
√ running red lights, speeding
? driving unregistered vehicles
?? driving while unlicensed
Copyright
8
2008-09
'Blacklist in Camera' ANPR Architecture
Tightly-
Coupled
Processing
Camera
& OCR
Alerts
Police
Cars
Copyright
9
2008-09
'Blacklist in Camera' ANPR Architecture
Tightly-
Sources of
Coupled Data-Sets
Pro cessing
Camera
& OCR
Alerts
Police
Cars
Copyright
10
2008-09
'Blacklist in Camera' ANPR Architecture
Tightly- Sources of
Coupled Data -Sets
Processing
Camera
& OCR
Alerts
Alerts Only
Alerts
Police Operational
Cars Policing
Copyright
11
2008-09
ANPR Quality
• Alliances of purveyors and purchasers suggest that
data extraction is accurate and reliable ... BUT ...
• Very little evidence is publicly available
• There appear to be no independent tests
• Many factors reduce reliability, including:
• the nature and condition of the registration plates
• the condition of the camera lens
• the conditions of the light-path and back-lighting
• The extraction is by its nature 'fuzzy',
and confidence threshholds have to be set
• Reliable extraction of the registration data may be
as low as 70% even under favourable conditions
Copyright
12
2008-09
ANPR Traffic Applications
Some Implications
• Deterrence of Targeted Behaviour
• Targeted Fines and Points Deductions
• Substantial Resources Required,
in particular Police Cars Downstream
• False-Negatives Escape
• False-Positives Suffer:
• Financial Impacts
• Licence-Retention Impacts
Copyright
13
2008-09
LEAs – Operational Applications
• Detection and Interception of:
• Wanted Vehicles, in particular:
• 'Reported Stolen'
• 'Get-Away Cars'
• Vehicles associated with Persons of Interest
• Dependent on:
• Real-Time Acccess to ...
• Real-Time-Maintained Data Sources
Copyright
14
2008-09
LEA Operational Applications
Quality Factors and Implications
• Data-Source Quality Factors are critical,
esp. Accuracy, Precision and Currency
• (Rare?) Instances of Large Benefits
• (Common?) Instances of Error:
• High Risk to Vehicle Occupants
because of the Interceptor's Suspicions
• Substantial Embarrassment, Confusion
• Likelihood of Collateral Police Actions
– arbitrary vehicle inspection, search
Copyright
15
2008-09
LEAs – Intelligence Applications
• Retrospective Analysis of Vehicle Movements:
• Detection of Duplicates
• False Registration Numbers
• Retrospective Inferences about
Owner Location and Movements
• Retrospective Inferences about
Co-Location, and Co-Location Frequency, of:
• Vehicles
• People
• Real-Time Inferences about Location, Co-Location
Copyright
16
2008-09
Mass Surveillance ANPR Architecture
Camera
& OCR
All
Sightings
Alerts
Police Operational
Cars Policing
Copyright
17
2008-09
Mass Surveillance ANPR Architecture
Central
Camera Processing
& OCR & Storage
All All
Sightings Sightings
Alerts
Police Operational
Cars Policing
Copyright
18
2008-09
LEAs – Intelligence Applications
Quality Factors
• Unreliable Extraction of Registration Data
• Data Collection Speculative
i.e. without Due Cause / Reasonable Grounds for Suspicion
This protection is a foundation of a free
society
• Retention Periods unclear and possibly very long
• Use of Probabilistic (Speculative) Data Mining
in order to generate suspicions
Copyright
19
2008-09
ANPR Deployments in Australia
• In most States and Territories, one or more agencies
has deployed or at least piloted ANPR
• 300-400 cameras acquired, some currently
operational
• One longstanding application exists:
• NSW RTA Safe-T-Cam for trucks
• 24 fixed-location cameras since 1989
• relatively recently migrated to ANPR
Copyright
20
2008-09
Features of
ANPR Deployments in Australia
• Every Single Deployment Lacks:
• Explicit Legal Authority
• Public Justification
• Public Information
• Public Consultation
• Operational Transparency
• Effective Regulatory Control
• Effective Privacy Laws
Copyright
21
2008-09
Submissions expressing serious concern about
privacy:
• APF
• OFPC
• OVPC
• QCCL
Copyright
22
2008-09
• OVPC: "The whole concept of an individual’s right to
anonymity is sacrificed: it is no longer possible to drive on a
public road anonymously, even if one is doing nothing wrong"
• OFPC: "ANPR can result in the routine collection of the personal
information of large numbers of people. For many of these people,
there may be no cause for suspicion and hence no reason to
collect information about them. A widespread ANPR system may
permit government agencies to track a large number of vehicles
(and individuals), revealing where individuals have been, when and
potentially with whom. Other than in specific circumstances, this
does not seem to be information that government agencies would
routinely need to know about members of the community ... The
Office would caution against establishing infrastructure that
could [be] used in such an expansive and invasive manner"
Copyright
23
2008-09
Recommendations of the Qld Parliamentary Committee:
• [because there is no current justification,]
further research of the road safety benefits of ANPR
• [because the proposal is so privacy-intrusive,]
crucial legislative safeguards ... to protect ... privacy
• [because quality is low,]
the resolution of technical problems that prevent ANPR
devices reading some number plate designs
Copyright
24
2008-09
Coordinative Activities
by Crimtrac
• The national LEA information systems operator
(e.g. fingerprint, DNA databases)
• Given $2.3m for an 'ANPR Scoping Study' 2007-08
• Privacy Issues Analysis conducted Nov 2007
• "We have not yet determined exactly the extent to
which we would capture all data. It may well be
that we only capture hot list data"
(Transcript of Evidence to Qld Parltry Travelsafe
Committee, 14 Mar 2008, p. 17)
• PIA and Consultation (Jun-Nov 2008)
Copyright
25
2008-09
Crimtrac's PIA Consultation Paper
June 2008
" ... the system will collect and store ... all sightings of all vehicle
passengers"
A 'National Automated Vehicle Recognition System' (NAVR)
"data-matching to identify alerts would take place centrally ..."
"sightings would be collected for all vehicles passing a camera site,
and would contain an overhead image of the vehicle at sufficient
resolution so that the driver or passenger could be identified if
appropriate
"[from] 300 fixed and 100 mobile to 4000 fixed and 500 mobile
cameras"
"all ANPR data would be held for five years"
an indicative 70 million sightings per day – implying 127 billion
photographs and associated metadata over a rolling 5-year cycle
Copyright
26
2008-09
Crimtrac's Untrustworthiness
• The position established in May 2008 is
inconsistent with the statements of mid-Mar 2008
• Committed to Mass Surveillance ANPR
• Expressly Facilitative of Mass Surveillance
• No Consideration of the negative consequences
• PIA Report withheld, despite an
understanding it would be published
• Scoping Study Report withheld
Copyright
27
2008-09
Covert Implementation of Mass Vehicle Surveillance
Conclusions
• LEAs are implementing ANPR covertly
i.e. without full public information,
without oversight, without express authority
• LEAs are using Mass Surveillance ANPR,
not Blackist-in-Camera architecture
• Crimtrac is implementing the facilitative
mechanism for Mass Surveillance ANPR
• After initially adopting some degree of
openness, Crimtrac is operating covertly
Copyright
28
2008-09
Covert Implementation of Mass Vehicle Surveillance
Implications
For LEAs
• A further step in the slide into untrustworthiness
• Greatly increased risk of behaviour above the law
• Greatly increased risk of serious public distrust
For Australian society
• A profound reduction in civil liberties
• A groundbreaker for a surveillance society
• A major contributor to social breakdown
and anarchic behaviour
Copyright
29
2008-09
Covert Implementation of Mass Vehicle Surveillance
Policy Implications
• ANPR is a litmus test of the Rudd Government's
capacity to withstand the backroom pressure
put on it by the law enforcement community
• The Australian public wants law enforcement
agencies to have appropriate technology and
appropriate powers ... but not to the extent
that freedoms and democracy are undermined
Copyright
30
2008-09
Counterveillance Principles
1. Independent Evaluation of Technology
2. A Moratorium on Technology Deployments
3. Open Information Flows
4. Justification for Proposed Measures
5. Consultation and Participation
6. Evaluation
7. Design Principles
1. Balance
2. Independent Controls
3. Nymity and Multiple Identity
8. Rollback
Copyright
31
2008-09
The Covert Implementation
of Mass Vehicle Surveillance in Australia
Roger Clarke
Xamax Consultancy, Canberra
Visiting Professor at ANU, UNSW, and the Uni. of Hong Kong
Chair, Australian Privacy Foundation
http://www.anu.edu.au/Roger.Clarke/....
..../DV/ANPR-Surv {.html,.ppt}
Social Implications of Covert Policing Workshop – 7 April 2009
Copyright
32
2008-09
