Automated proximity discovery of networked cameras
Automated discovery of the relative positioning of a network of cameras that view a physical environment. The automated discovery is based on comparing TimeLines for the cameras. The TimeLines are time-stamped data relating to the camera's view, for example a sequence of time stamps and corresponding images captured by a camera at those time stamps. In one approach, the relative positioning is represented by a proximity graph of nodes connected by edges. The nodes represent spaces in the physical environment, and each edge between two nodes represents a pathway between the spaces represented by the two nodes.
1. A method implemented on a computer system for discovering a relative positioning of a network of cameras that view a physical environment, the method comprising:
receiving TimeLines for the cameras, each TimeLine for a camera comprising a sequences of time stamps and data relating to the camera's view at those time stamps, the data including images captured by the camera at those time stamps;
comparing the TimeLines captured by the cameras;
determining a proximity of a plurality of spaces viewed by the cameras, based on the comparison of the TimeLines, wherein the spaces are defined by the views of the cameras; and
constructing a proximity graph of the proximity of the spaces viewed by the cameras, the proximity graph comprising nodes connected by edges, the nodes representing the spaces defined by the views of the cameras and each edge between two nodes representing a pathway between the spaces represented by the two nodes.
2. The computer-implemented method of claim 1 wherein comparing the TimeLines captured by the cameras comprises:
identifying an object in the images captured by the cameras; and
tracking the object through the spaces viewed by the cameras, wherein the proximity graph is constructed based on tracking the object through the spaces viewed by the cameras.
3. The computer-implemented method of claim 2 wherein:
the object is tracked as moving without delay from viewed by a first group of one or more cameras to viewed by a different second group of one or more cameras;
the spaces viewed by the first and second groups of cameras are represented by existing first and second nodes in the proximity graph; and
constructing the proximity graph comprises adding an edge between the existing first node and the existing second node.
4. The computer-implemented method of claim 2 wherein:
the object is tracked as moving from viewed by a first group of one or more cameras to viewed by a different second group of one or more cameras, but with a delay during which the object is not viewed by any camera;
the spaces viewed by the first and second groups of cameras are represented by existing first and second nodes in the proximity graph; and
constructing the proximity graph comprises adding a new node for a non-viewed space, adding an edge between the existing first node and the new node for the non-viewed space, and adding an edge between the new node and the existing second node.
5. The computer-implemented method of claim 2 wherein:
the object is tracked as moving from viewed by only a first camera to viewed by both a first and a second camera;
the space viewed by the first camera is represented by an existing first node in the proximity graph, but no node in the proximity graph represents the space viewed by both the first and second cameras; and
constructing the proximity graph comprises adding a new node representing the space viewed by both the first and second cameras, and adding an edge between the existing first node and the new node.
6. The computer-implemented method of claim 1 wherein every camera in the network is represented by at least one node in the proximity graph.
7. The computer-implemented method of claim 1 wherein spaces viewed by two different groups of cameras are represented by two different nodes in the proximity graph, even if some cameras are in both groups.
8. The computer-implemented method of claim 1 wherein a space viewed by only one camera is represented by a different node than a space viewed by that one camera plus one or more other cameras.
9. The computer-implemented method of claim 1 wherein the proximity graph includes nodes that represent spaces that are not viewed by any camera in the network.
10. The computer-implemented method of claim 1 further comprising:
analyzing an image captured by a camera to determine a type of space viewed by that camera, wherein the proximity graph is constructed further based on the types of spaces viewed by cameras.
11. The computer-implemented method of claim 1 wherein the network of cameras views rooms within a building, the method further comprising:
analyzing an image captured by a camera to identify entries and exits to rooms viewed by that camera, wherein the proximity graph is constructed further based on the entries and exits viewed by cameras.
12. The computer-implemented method of claim 1 wherein a synchronization service receives the TimeLines and constructs the proximity graphs.
13. The computer-implemented method of claim 11 wherein the synchronization service receives the TimeLines via an API.
14. The computer-implemented method of claim 1 wherein the proximity graph is implemented as an array of objects.
15. The computer-implemented method of claim 13 wherein the array of objects includes a list of camera objects, each camera object representing one of the cameras in the network of cameras.
16. The computer-implemented method of claim 13 wherein each camera object includes a list of neighboring cameras.