Upload
esri
View
1.233
Download
25
Tags:
Embed Size (px)
DESCRIPTION
Presentation by Sanjay Rana from Blom Group on Esri European User Conference 2011.
Citation preview
Isovist Analyst 2.0An ArcMap 10 Add-In for Urban Viewshed Analyses : Visual SurveillanceAnalyses : Visual Surveillance
Sanjay RanaIdeas and Improvement Group @ BLOM (UK)
Topics�Brief Overview of Urban Viewshed Analysis and its
application in Visual Surveillance
� Isovist Analyst 2.0
2/ 25
� Isovist Analyst 2.0
�Visual Surveillance Planning and Evaluation using Isovist Analyst 2.0
3/ 25
Urban Viewshed or Isovist
Urban Viewshed Or IsovistDefinition� An urban viewshed or isovist is (typically a 2D) area visible
from a location in an indoor or outdoor urban environment.
Purpose
4/ 25
Purpose� Evaluating “Visual” Accessibility of Open Spaces is vital for
various applications e.g. visual surveillance, evacuation, crime pattern analyses, pedestrian wayfinding, and so on.
Example of an Isovist� Tate Britain Gallery, London, UK
Urban Viewshed Or IsovistBrief History� Proposed by Tandy (1967) and popularised by Benedikt
(1979) for studies in architecture, urban design, urban planning, transport planning.
5/ 25
� Research on this theme now come under the field of “Space Syntax”.
� Relevant Wikipedia Entries: isovist space syntax visibility graph analysis space network analysis software
Urban Viewshed Or IsovistProperties of Isovists� Geometrical measures
� Shortest/Longest line of sights� Area� Shape Ratios
6/ 25
� Shape Ratios
� Topological measures� Clustering Coefficient� Control
r d
Urban Viewshed Or Isovist
Dense Mesh of 4298 Observers
Isovist Area
High
Tate
Gal
lery
, Lo
nd
on
7/ 25
Maximum Diametric Length
CircularityLow
Tate
Gal
lery
, Lo
nd
on
Urban Viewshed Or IsovistSome Urban Viewshed Scenarios in Homeland Security:
8/ 25
Visual Surveillance Wayfinding in Evacuation Discarded Needles in City of Montreal
9/ 25
Visual Surveillance
Visual SurveillanceTwo types of visual surveillance� Passive or Natural Visual Surveillance
� Design of urban open spaces so as to ensure that occupants can self-regulate themselves and reduce the fear of crime.
10/ 25
fear of crime.� Cheap, long lasting but involves a risk.
� Active or Artificial Visual Surveillance� Direct and intrusive remote monitoring via devices such
as CCTV or human observers.� Potentially more effective but costly.
11/ 25
Passive Visual Surveillance
Passive Visual SurveillancePassive Visual Surveillance
� The aim is to create the perception of being observed or feeling safe by suitable architectural design of an open space:
� Occupants can see other occupants and entry/exist points in
12/ 25
� Occupants can see other occupants and entry/exist points in space, so that they can self-regulate themselves, e.g., in public spaces,
� Occupants always feel being watched by an invisible observer.
� Measures include removal of blind spots, dark alleys, installing one-way screens (Venetian blinds) etc.
Passive Visual SurveillancePassive Visual Surveillance� Perception of safety and well-being in an open space can be
correlated to the visibility properties of the open space for instance, poorly lit (e.g., by natural light) areas generate a fear of crime.
13/ 25
� People follow cues (e.g., long corridors) in the space while exploring open spaces.
� Isovist measures are correlated to empirical data on perception of crime and people behaviour to design new urban open spaces.
Passive Visual SurveillanceNatural Visual Surveillance
14/ 25
Greene, M., and Greene, R., 2004, Urban safety in residential areas, Space Syntax Conference.
Passive Visual Surveillance
� A central guard tower, surrounded by prison cells but the prisoners can neither see each other nor the
Famous Example of Passive Visual Surveillance: The Panopticon Prison Design by Jeremy Bentham
15/ 25
neither see each other nor the prison guards therefore creating the perception of being under constant invisible observation.
� Examples include Twin Towers Correctional Facility (Los Angeles) and Pentonville Prison (England).
16/ 25
Active Visual Surveillance
Active Visual SurveillanceActive Visual Surveillance� The use of visual surveillance has dramatically increased in
recent years, as demands arose for monitoring,� Criminal behaviour in public space,� Traffic,
17/ 25
� Traffic,� etc. etc.
� Several thousand millions of pounds are spent on sales and rental of CCTV and other types of visual surveillance.
Active Visual Surveillance
18/ 25
CCTV Cameras in Manhattan
Active Visual SurveillanceActive Visual Surveillance
� The aim is to ensure complete/maximum visual coverage of an open space.
� It involves an iterative, manual and gut-feel process of
19/ 25
� It involves an iterative, manual and gut-feel process of trying various layouts until a satisfactory solution has been found.
� Traditionally done by architects, security consultants and urban planners using CAD software.
Active Visual SurveillanceExample Of Active Visual Surveillance
Identification of the Four Suspects In London Transport Network
20/ 25
Transport Network Bombings (July, 2005)
Active Visual SurveillanceActive Visual Surveillance� Challenges:
� Costs of installation and maintenance, and coverage of CCTV cameras need to be addressed.
21/ 25
� The problem of ensuring maximum visual coverage with a minimum number of CCTV/observers is broadly similar to Location-Allocation, Chinese Postman/Travelling Salesmanproblems in GIS, and most importantly closely similar to the Art Gallery Problem in Computer Science.
22/ 25
Art Gallery Problem
Active Visual SurveillanceArt Gallery Problem (AGP)� Proposed by Victor Klee in 1973.� What’s the minimum number of guards necessary to provide
complete visual coverage of an art gallery?� Solution of the AGP is going to be relevant for an affordable
23/ 25
� Solution of the AGP is going to be relevant for an affordable and effective CCTV network.
� However, there are no robust algorithms that guarantee a general purpose solution.
� Mathematically, the problem is NP-hard. Thus, it is analytically and computationally non-trivial and unsolvable at present therefore approximate solutions are used.
Active Visual SurveillanceArt Gallery Problem� Most existing algorithms assume a certain topology of the art
gallery e.g., rectilinear, convex. A well known formula to solve the AGP is by Chv’atal (1975),
24/ 25
For a gallery with n walls and h holes, at most (n+h/3) number of guards are sufficient to provide complete visual coverage,
� This solution doesn’t guarantee the minimum number in all shapes but definitely yields a number enough for complete visual coverage e.g.
How Many Guards Are Sufficient to
Cover the Gallery ?
71 Walls + 0 Holes
(n+h)/3 = 23?
9
25/ 25
26/ 25
Rank and Overlap Elimination (ROPE)
Active Visual SurveillanceRank and Overlap Elimination (ROPE) For Solving AGP� It is essentially a greedy-search method.
� The open space is discretised with a dense mesh of potential observers.
27/ 25
observers.
� Each observer is assigned a rank based on the number of other observers seen by it. Higher rank means higher visibility.
� Starting with the highest ranking observer (observer with largest isovist area), an iterative process selects the optimal observers
Active Visual SurveillanceSTART
Pick highest ranking observer remaining in potential observers
Remove all potential observers ROPE Selection
28/ 25
Remove all potential observers visible to highest ranking observer
Potential observers left in open space ?
STOP
No
Yes
ROPE Selection Process Flowchart
Active Visual Surveillance
29/ 25
Dense Mesh of Potential Observers
Isovist of the highest ranking observer
Two optimal observers
Active Visual Surveillance
Optimal Observers
ROPE technique doesn’t guarantee the minimum number of observers in all shapes.
30/ 25
Optimal ObserversROPE Observers
But it’s not worse than the analytical solution (Chv’atal, 1975):
(10 walls + 0 holes) / 3 = 3 guards
31/ 25
Isovist Analyst 2.0
Isovist Analyst 2.0� ArcMap add-in for computing isovist and over 20 isovist
properties.� 1.x versions released about a decade ago as extensions to
ArcView 3.x (Screen shot):� Still the only such application embedded inside ESRI suite of
32/ 25
� Still the only such application embedded inside ESRI suite of GIS. Used worldwide for research and teaching purposes.
� Voted Second Best User Software Application at 2007 ESRI International User Conference.
� Open space can be of arbitrary topology (i.e., with/out holes) and geometry (i.e., lines, polygons).
� Key achievement is the integration of urban visibility analyses in GIS.
Isovist Analyst 2.0How it works� Technique of Ray Tracing is used to cast rays (line of sights)
at a user-defined angular interval into the open space from a viewpoint.
� Nearest points of ray intersection with obstacles are
33/ 25
� Nearest points of ray intersection with obstacles are connected together to form the isovist polygon.
Requirements� Point Feature Layers as Observers/Viewpoints.� Polyline or Polygon Feature Layers as Obstacles.
Isovist Analyst 2.0Software Components:
� Combination of superior cartographic visualisation and data interactivity of ArcMap 10 with light-weight and easy to code Open Source geospatial libraries.
�ArcObjects
34/ 25
�ArcObjects�NetTopologySuite, MapTools, GeoAPI
35/ 25
Isovist Analyst 2.0 Demonstration
Isovist Analyst 2.0Demonstrations:� Tate Gallery Isovist
� Data Used: Tate Gallery Outline. Tate Gallery Demo
� A (very) crude evaluation of CCTV Coverage of Norwich City
36/ 25
� A (very) crude evaluation of CCTV Coverage of Norwich City Centre, UK.� Data Used:� Blom3D Level of Data 1 (LOD 1) Building Footprint polygons as
Obstacles� Blom WebViewer for visualising surveillance context e.g.
Landuse
Planning and Evaluating Visual Surveillance using IA 2.0� Disclaimer: Solely to demonstrate the application context and
doesn’t in any way indicate any actual issue being studied.
� Norwich is an important city in East of England.
37/ 25
� City was basically chosen because CCTV and Crime data was found on the internet, and BLOM contains good coverage of the city in aerial photographs and vector dataNorwich
Planning and Evaluating Visual Surveillance using IA 2.0Current distribution of CCTV in Norwich City Centre.Source: http://sites.google.com/site/norwichpolicecctv/, last accessed 23rd October 2011.
38/ 25
7288 Norwich City Centre Street Crimes
66% uncovered
Planning and Evaluating Visual Surveillance using IA 2.0
Where are the CCTVs and street crimes?
6400 Potential CCTV Locations
717 for full coverage
46 Sites after a greedy search
39/ 25
Violent crime
Vehicle crime
Robbery
Other crime
Burglary
Anti-social behaviour
Street Crimes Dec 2010 - Aug 2011
ConclusionsConclusions and Future DirectionsThis work demonstrates� Visual Surveillance has become a necessary aspect of
modern society,� Planning of Visual Surveillance is essentially a geospatial
40/ 25
� Planning of Visual Surveillance is essentially a geospatial exercise.
� Most GISs do not provide in-built functions for planning visual surveillance.
� Automated algorithms for evaluating the structure of open spaces and planning visual coverage, in relation to planning visual surveillance can be easily developed in a GIS.
Disclaimer and Credits� All views and work expressed are solely my own. Blom has
no responsibility for any errors or opinions.� Images:
� http://www.chelmsford.gov.uk/media/image/m/d/Monitor_wall_%28o%29_large.jpg
41/ 25
8o%29_large.jpg� http://urbandesign.tfl.gov.uk/Design-Guidance/London-
Rail/Crossrail/Underground-Rail-%281%29/Routes-%281%29.aspx
� BLOM for access to building footprints and Blom WebViewer data� www. isee.com for the CCTV coverage of Manhattan image.� Conroy and Dalton (1999) for the Tate Gallery outline.� www.police.uk for crime data
43/ 25
Isovist Analyst 1.x
44/ 25