A widely-applied fashion of security surveillance

also known as closed circuit television（CCTV）

in micro-spatial environments

such as inside a building

is to have security personnel sit in front of a wall of monitors

which corresponds to spatially-distributed video cameras

and respons（e.g.alarms

access control

etc.） on any emergent events or suspicious objects.In particular

if a watched object is mobile

e.g.a theft

security personnel need to keep track of the object’s movement through frequently switching their visual focus among monitors.Even for experienced security personnel

the tracking process is flustered if the number of monitors is large or the monitored object moves fast.To the best of our knowledge

there is still lack of approaches which can provide security personnel with leads to track and estimate the movement of monitored objects in a micro-spatial environment.In view of this

we present a GIS-based approach to assist video surveillance indoor.In the proposed approach

all accessible places in a building and the locations of video cameras are represented with a network data model.Generally

monitored objects can only stay or move in accessible places of a building

e.g.rooms

exits（e.g.gates

stairs

elevators

etc.）

and hallways.These accessible places in a building can be represented with a node-arc network data model.The network consists of nodes and arcs.Nodes

corresponding to rooms or exits

are end points of an arc

and arcs

corresponding to hallways

can only meet with each other at a node.Monitored objects can only access to a node（i.e.room or exit） through arcs（i.e.hallways）.With this network representation

the topological relationships among accessible places can be clearly illustrated

and network analysis can be employed to calculate the accessibility in a building.Instead of linear hallway

in some cases

rooms or exits are linked or accessed to through a lobby.Although a lobby can be also treated as a room

one node in the geometric center of the lobby is not enough to reflect the walking behavior of people within it.Therefore

additional arcs are needed to fill in the spacious area of a lobby.Our approach is to link the geometric center and associated rooms or exits with arcs

and then link the middle points of these arcs with additional arcs.The above procedure can represent the accessibility in a single floor.To represent the accessibility of a multi-level building

we can link stair or elevator exits of different floors with arcs.Each camera is represented as a point identical with a node or along an arc closest to the camera’s location in a node-arc network and located with linear-referencing system（LRS）

which is a well-known data model in GIS for transportation（GIS-T）.A monitor corresponding to a camera generally displays the field of vision（FOV） through this camera.For each camera

we can delineate a FOV on the node-arc network.Depending on the installation gradient of a camera

its FOV may surround this camera or not.In this paper

we suppose each camera is surrounded by its FOV.The above definition facilitates us encoding topological relationships between one camera’s FOV and nearby objects including other cameras’ FOV and rooms in order to provide security personnel with leads of tracking monitored objects.For each camera

8 attributes are employed to record directly-linked cameras and rooms at 4 directions.Based on the network model and the data structure

a number of applications can be realized.One of them elaborated in the paper is to locate suspicious moving in a short time objects.Besides the procedure

a detailed description is given to explain how to implement the procedure and how to link the above research output to monitors and cameras.