CROSS-REFERENCES TO RELATED APPLICATIONS
[1] None.
FIELD OF THE INVENTION
[2] The invention generally relates to surveillance systems and in particular to a
method of maximizing camera coverage in a surveillance system having a plurality of cameras.
DESCRIPTION OF RELATED ART
[3] Intelligent video surveillance has become an important research area that is
supported with the advancements in camera technology and the demand for security. Optimal camera placement is essential for better surveillance and coverage. The efficient placement of surveillance cameras has an enormous impact on the performance and cost of the surveillance system. The objective of the camera placement problem (coverage problem) is to maximize the coverage with minimum number of cameras. Despite its wide applicability, the placement of cameras in a surveillance room is an NP hard problem. Therefore, optimization techniques are used to obtain a near optimal solution for the camera placement problem.
[4] The optimal selection and placement of cameras in a multi-camera system is
another important factor. The selection of optimal camera configurations (camera locations, orientations, etc.) for better surveillance is a challenging problem. In surveillance applications, the coverage is one of the priority objectives in practice and therefore the most studied field in recent literature. Various research problems related to

camera coverage are: (i) determining optimal camera poses with respect to maximizing coverage while camera locations are predetermined (ii) determining camera location and pose when the total number of cameras are fixed (iii) determining minimum number of cameras, their position and orientation for a minimally required percentage of coverage that meets the coverage constraint.
[5] Surveillance networks have a wide range of commercial and military
applications from video surveillance to smart home and from traffic monitoring to antiterrorism. Some modern visual surveillance systems require automatic persistent coverage and demand multiple camera coverage over the high secure area. For applications such as object tracking, a sufficient overlap between cameras field of view is required for successful tracking of object. Some applications require coverage overlapping to satisfy security needs when one camera failed. Whereas some applications constraint coverage overlapping to maximize total coverage gain of the visual sensor network. Therefore, the optimization algorithms must be designed to achieve an optimal balance between certain coverage gain and coverage overlap to organize a multi-viewing surveillance system.
[6] Some of the algorithms used for optimization tasks include Greedy
optimization methods, Integer linear programming (ILP) approach and genetic algorithm (GA). In a multi sensor system, the placement of sensors is based on the application requirement to achieve better surveillance. For security surveillance applications, maximum possible areas are covered with the available resources.
[7] The patent US9942468B2 discloses technologies for automatically
optimizing camera placements, numbers, and resolution in multi-camera monitoring and surveillance applications. "Optimal camera planning under versatile user constraints in multi-camera image processing systems," IEEE Trans. Image Process (Liu et al, 2014)

proposes an optimal camera planning based on trans-dimensional simulated annealing algorithm. The algorithm deals the scalability issues with large scale networks compared to other state-of-the-art methods. Sometimes a fixed static configuration of camera network is not sufficient for sensing applications. "Dynamic reconfiguration in camera networks: A short survey", IEEE Transactions on Circuits and Systems for Video Technology (Piciarelli, 2015) discloses that dynamic network reconfiguration helps to optimize the network performance to the currently required specific tasks while considering the available resources. Due to high security demands, some applications require a secondary coverage. Most of the existing algorithms do not generalize well for large scale networks. There arises a need for optimizing methods that may perform well in large scale surveillance network.
[8] Disclosed herein are systems and methods that may identify optimal camera
configuration for a multi-camera network with predefined camera locations.

[9] In various embodiments a method of identifying optimal camera
configurations for maximum coverage of a surveillance area is disclosed. The surveillance area has a plurality of cameras fixed at predetermined locations, each of the plurality of cameras capable of being oriented at a plurality of predetermined orientations. The method includes reading a coverage value of each of the plurality of cameras set at a first orientation, by a receiving module of a computing system. The coverage value of each of the plurality of cameras indicating an area of coverage of the surveillance area. A total coverage value is calculated from the coverage values obtained at the first orientation of each of the plurality of cameras by a processing module of the computing system. The total coverage value is stored as a maximum coverage value in a memory module of the computing system.
[10] In various embodiments one of the plurality of cameras is moved to a next
orientation in the plurality of predetermined orientations of the cameras, by a control module of the computing system. The coverage value of each of the plurality of cameras is read, by the receiving module of the computing system. A total coverage value is calculated from the coverage value obtained from each of the plurality of cameras by the processing module of the computing system. The maximum coverage value is compared with the calculated total coverage value by a processing module of the computing system. The larger of the compared total coverage values is stored as the maximum coverage value in the memory module of the computing system and an orientation corresponding to the stored total coverage value is selected, for the one of the plurality of cameras. In various embodiments the above steps are iterated for each of the plurality of cameras at the next orientation and through the plurality of predetermined orientations and the maximum coverage of the surveillance area is obtained.

[11] In various embodiments a total coverage gain is given by
o [ L M N
cL = 2_i max 2J 2J 2J co,i,ij © CL-I
0=1 z=it=iy=i
where CL coverage gain, O is the orientation and L is the location of cameras.
[12] The invention in various embodiments discloses a method of identifying
optimal camera configurations for maximum coverage of a surveillance area monitored using a surveillance system. The surveillance system includes a plurality of cameras fixed at predetermined locations and orientations and a computing system having a receiving module, a processing module, a control module and a memory module. The method includes the steps of dividing the surveillance area into one or more grids by the processing module. A vote value provided to the one or more grids by the processing module based on a percentage of coverage by a first camera at a first orientation, and also on whether the grid has already been covered by any camera configured previously at an orientation is read and stored in a memory module. The vote value is provided for the plurality of cameras fixed at predetermined locations and all predefined orientations of the plurality of cameras.
[13] An accumulated vote value is calculated for each grid by adding the vote
values for each grid obtained. In various embodiments a vote value is provided to the cameras set at a position and orientation by the processing module, based on the accumulated vote value for each grid and also on whether the camera has already been covered previously by any grid. The vote value is read and stored in a memory module. The process is repeated iteratively for the one or more grids. An accumulated vote value is calculated for each of the plurality of cameras for all defined orientations by adding the vote values for each cameras and an orientation is selected for one of the plurality of cameras corresponding to the accumulated vote value that is maximum. The method is repeated iteratively through all defined orientations for each of the plurality of cameras and maximum coverage of the surveillance area is obtained.

[14] In one embodiment the vote value provided to the grids within the one or
more grids by the processing module is
if the grid has not been covered previously by any camera, set at an orientation, wherein Cki{Li, 6j) is the percentage of grid G^^ covered by the camera.
[15] In another embodiment the vote value provided to the grids within the one
or more grids by the processing module is
Vi,j(H(k,l)) = -Ck,l{Li,9j),
if the grid has already been covered previously by any camera, set at an orientation, wherein H^k^ is a grid that has already been covered previously by any camera, set at an orientation.
[16] The vote value provided to the cameras set at a position and orientation by
the processing module is
V(kAk.Qj) = a ^]) (r^ \~Pv„J) (u'l) \ > wnerein a and P are regularization
parameters where a controls a total coverage and /? controls an overlap ratio. In various embodiments the values of a and /? ranges between 0 and 1.
[17] In one embodiments when the value of a is in a range 0.5 to 1 the method
operates in a maximizing coverage mode wherein the cameras are set at orientations that cover the regions that have a lesser chance of being covered.
[18] In another embodiment when /? is in a range 0.5 to 1 and a is in a range 0 to
0.4 the method operates in a minimizing overlap mode wherein the cameras are set at orientations that reduce the coverage overlap.
[19] In some embodiments the surveillance area has one or more critical regions.
In various embodiments the minimum coverage threshold Tcov of the one or more critical regions is 100%.

[20] In various embodiments performing voting for a camera set at an orientation
at the critical region by the grids is weighted
where Wc is the weight factor for the vote from the grids. In various embodiments the critical region is covered by 2 or more cameras in the network. In various embodiments the critical region is covered by 'm' cameras, the coverage value associated with each grid is initialized with ' 1-m'.
where Gcov is the coverage value associated with each grid.
[21] In various embodiments a surveillance system is disclosed. The system
includes a plurality of cameras fixed at predetermined locations in a surveillance area and configured to be oriented at a plurality of predetermined orientations and a computing system connecting the plurality of cameras. The computing system includes a control module, a receiving module, a memory module and a processing module. The control module is configured to set the plurality of cameras to a first orientation and move each of the plurality of cameras one at a time to a next orientation in the plurality of predetermined orientations of the cameras. The receiving module is configured to read a coverage value of each of the plurality of cameras set at a first orientation and next predetermined orientations. The memory module is configured to store one or more values. In various embodiments the processing module is configured to calculate a total coverage value from the coverage values obtained at the first orientation of each of the plurality of cameras and store the total coverage value as a maximum coverage value in the memory module.
[22] In various embodiments the plurality of cameras are moved to a next
orientation in the plurality of predetermined orientations by initiating a command to the control module and the coverage value of each of the plurality of cameras (301-1, 301-2,...301-8) obtained from the receiving module (306) is read. A total coverage value is

calculated from the coverage value obtained from each of the plurality of cameras (301-1, 301-2,...301-8). The maximum coverage value and the calculated total coverage value are compared and the larger of the compared total coverage values is stored as the maximum coverage value. An orientation corresponding to the stored total coverage value is selected, for the one of the plurality of cameras. In various embodiments the processing module initiates a command to the control module until the orientations of the plurality of cameras are selected to obtain the maximum coverage of the surveillance area. In various embodiments the memory module is configured to store the maximum coverage value and the larger of the compared total coverage values.
[23] The invention discloses a surveillance system. The surveillance system
includes a plurality of cameras fixed at predetermined locations in a surveillance area (302) and configured to be oriented at a plurality of predetermined orientations and a computing system connecting the plurality of cameras over a network. The computing system includes a control module, a receiving module, a memory module and a processing module. The control module is configured to set the plurality of cameras to a first orientation and move each of the plurality of cameras one at a time to a next orientation in the plurality of predetermined orientations of the cameras. The receiving module is configured to read a coverage value of each of the plurality of cameras set at a first orientation and next predetermined orientations. The memory module is configured to store one or more values.
[24] In various embodiments a processing module is configured to divide the
surveillance area into one or more grids and read and store in a memory module a vote value provided to the one or more grids based on a percentage of coverage by a first camera and each of the plurality of cameras at a first orientation and at all predefined orientations, and also on whether the grid has already been covered by any camera configured previously at an orientation. An accumulated vote value is calculated for each grid by adding the vote values for each grid. A vote value provided to the cameras set at a position and orientation by each of the one or more grids based on the accumulated vote value for each grid and also on whether the camera has already been covered previously by any grid is read and store in the memory module. An accumulated vote value for each

of the plurality of cameras is calculated for all defined orientations by adding the vote values for each camera. In various embodiments an orientation is selected for one of the plurality of cameras corresponding to the accumulated vote value that is maximum. In various embodiments the processing module initiates a command to the control module until the orientations of the plurality of cameras are selected to obtain the maximum coverage of the surveillance area.
[25] In various embodiments the memory module is configured to store the vote
value provided to the one or more grids, the vote value provided to the cameras set at a position and orientation and the accumulated vote value. In one embodiment the processing module is configured to provide vote values to the grids within the one or more grids and is given by
if the grid has not been covered previously by any camera, set at an orientation, wherein Ckii(Lt, 6j) is the percentage of grid G^Qcovered by the camera.
[26] In one embodiment the processing module is configured to provide vote
values to the grids within the one or more grids and is given by
if the grid has already been covered previously by any camera, set at an orientation, wherein H^k^ is a grid that has already been covered previously by any camera, set at an orientation.
[27] In various embodiments the processing module is configured to provide vote
value to the cameras set at a position and orientation given by
wherein a and /? are regularization parameters where a controls a total coverage and /? controls an overlap ratio. In various embodiments the system is configured to cover one or more critical regions by 2 or more cameras.

[28] The invention has other advantages and features which will be more readily apparent from the following detailed description of the invention and the appended claims, when taken in conjunction with the accompanying drawings, in which: [29] FIG. 1A illustrates a method of identifying optimal camera configurations. [30] FIG. IB illustrates the execution orders of the algorithm on a binary map image with 16 fixed cameras locations, each with 8 possible camera orientations. [31] FIG. 2 illustrates a grid based method of optimizing camera coverage. [32] FIG. 3 A shows a diagrammatic representation of the grid-based algorithm. [33] FIG. 3B shows the execution orders of the grid-based algorithm. [34] FIG. 4 depicts a surveillance system having a plurality of cameras and a computing system.
[35] FIG. 5A illustrates modelling the coverage area of a visual surveillance system and a triangular coverage area.
[36] FIG. 5B illustrates modeling coverage overlap between two cameras in the surveillance system.
[37] FIG. 6A illustrates the simulated map image corresponding to the placement of cameras.
[38] FIG. 6B illustrates the simulated results of the Local Greedy (LG) Method. [39] FIG. 6C illustrates the simulated results of the Global Greedy (GG) Method. [40] FIG. 6D illustrates the simulated results of the Alternate Global Greedy (AGG) Method.
[41] FIG. 6E illustrates the simulated results of the Greedy Grid-Voting algorithm-maximum coverage mode (GGV-max).
[42] FIG. 6F illustrates the simulated results of the Greedy Grid-Voting algorithm-minimize overlap mode (GGV-min).

[43] FIG. 7A illustrates the coverage results at each iteration in GGV- max mode. [44] FIG. 7B illustrates the coverage results at each iteration in GGV- min mode. [45] FIG. 8A shows the total coverage area against the number of configured cameras, across the different optimization techniques LG, GG, AGG, GGV-max and GGV-min method.
[46] FIG. 8B shows the overlap area against the number of configured cameras, across the different optimization techniques LG, GG, AGG, GGV-max and GGV-min method. [47] FIG. 9A illustrates heatmap representing total coverage with different combinations of regularization parameters.
[48] FIG. 9B illustrates heatmap representing overlap area at different combinations of regularization parameters.
[49] FIG. 10A shows the simulation image of an examination hall. [50] FIG. 10B shows the total area covered by the LG algorithm. [51] FIG. IOC shows the total area covered by the GG algorithm. [52] FIG. 10D shows the total area covered by the AGG algorithm. [53] FIG. 10E shows the total area covered by the GGV-max coverage algorithm. [54] FIG. 10F shows the total area covered by the GGV-min overlap algorithm. [55] FIG. 11A illustrates the total coverage obtained in an examination hall before setting priority to the critical regions.
[56] FIG. 11B illustrates the total coverage obtained in an examination hall after setting priority to the critical regions.
[57] FIG. 11C illustrates the comparison chart of total coverage at critical regions in the examination hall.
[58] Referring to the drawings, like numbers indicate like parts throughout the
views.

[59] While the invention has been disclosed with reference to certain embodiments, it will be understood by those skilled in the art, that various changes may be made and equivalents may be substituted without departing from the scope of the invention. In addition, many modifications may be made, to adapt to a particular situation or material to the teachings of the invention, without departing from its scope.
[60] Throughout the specification and claims, the following terms take the
meanings explicitly associated herein unless the context clearly dictates otherwise. The meaning of "a", "an", and "the" include plural references. The meaning of "in" includes "in" and "on." Referring to the drawings, like numbers indicate like parts throughout the views. Additionally, a reference to the singular includes a reference to the plural unless otherwise stated or inconsistent with the disclosure herein.
[61] The invention in its various embodiments discloses methods of maximizing
coverage of a surveillance area in a surveillance system. The method includes identifying optimal configurations for one or more cameras fixed at predetermined locations by conducting a search for the predetermined locations that provide maximum coverage at an orientation. A grid-based iterative voting method for coverage optimization is also disclosed. The grid-based method involves a two-step voting process that includes a forward voting and a reverse voting process. In the forward voting process each location orientation pair votes to the grids that comes under its coverage and in the reverse voting process each grid votes to the location orientation pairs that cover the grid.
[62] The method 100 of maximizing coverage of a surveillance area in a
surveillance system is illustrated in FIG. 1A. The surveillance system has a plurality of cameras placed at predetermined locations across the surveillance area. The cameras are

capable of being oriented at a plurality of predetermined orientations. The predetermined orientations may include orientation angles between 0° to 360°. Each camera is connected to a computing system that has a receiving module, a processing module, a memory module and a display module. Data from the plurality of cameras are transmitted to the computing system.
[63] The method 100 as shown in FIG. 1A includes setting up the plurality of cameras at a first orientation 9i in step 101. The area of coverage of each camera at the first orientation 9i is read by a receiving module in the computing system and stored in a memory module as a coverage value for each of the plurality of cameras. The area of coverage of each camera forms part of the surveillance area. In step 102 a total coverage value for the first orientation 9i is calculated by the processing module. The total coverage value for the first orientation 6i is calculated from the area of coverage of each camera at the first orientation 9i. In step 103 the total coverage value calculated for the first orientation 9i is stored in the memory module as a maximum coverage value Cmax- In step 106 the orientation of one of the cameras from the plurality of cameras is moved by a control module to a next orientation that is predetermined. The coverage value of each of the plurality of the cameras are read by the receiving module and the total coverage value Ctot is calculated in step 107. The maximum coverage value Cmax and the total coverage value Qot is compared by the processing module in step 108 and in step 109 the larger of Cmax and Ctot is stored as the maximum coverage value Cmax. The camera is set at the orientation corresponding to the stored maximum coverage value Cmax. The next camera is selected according to a predetermined order and is turned to the next orientation. FIG. IB illustrates the execution orders of the algorithm. Steps 106 to 110 are repeated for the next camera. Steps 106 to 110 are repeated iteratively for the plurality of cameras and through all the predetermined orientations.

[64] Each camera is set at the orientation in which the coverage of the surveillance area is maximum. In various embodiments if the predetermined number of orientations are 'n' the number of iterations performed for each camera location L;is n i.e. for each orientation the method may select optimal camera location. In various embodiments the method performs a search for the camera locations that provide maximum coverage at an orientation.
where CL coverage gain, O is the orientation, L is the location of camera. The coverage gain is the ratio of the surveillance area covered by the surveillance system to the total surveillance area. The method produces better coverage results with minimum overlap when compared to other greedy methods. Also the method is capable of selecting a location for a camera orientation in such a way that it maximizes the coverage.
[66] In various embodiments a grid-based method 200 of identifying optimal
camera configurations for maximum coverage of a surveillance area that is monitored using a surveillance system is disclosed. The surveillance system has a plurality of cameras placed at predetermined locations L;, where i=l, 2, 3,.. .m across the surveillance area. The method 200 as shown in FIG. 2 includes dividing the surveillance area into one or more grids by the processing module in step 201. Each grid is of predetermined size. Each camera at location L;, where i=l, 2, 3,...m may take possible orientations 9j where j=l, 2, 3,....n. The camera at location L; and orientation 9j is considered as a location orientation pair (L;; 9j). In step 202a vote value provided to the one or more grids is read and stored in a memory module by the processing module. This is based on a percentage of coverage of the grids by a first camera at a first orientation, and also on whether the grid has already been covered by any camera configured previously at an orientation. In

various embodiments a forward voting process is perform in step 202 by each location orientation pair (L;, 9j) for the one or more grids G^k ^ that is covered by the location
orientation pair (L;, 9j). A vote value is provided for the grids based on a percentage of coverage of the grid by the location orientation pair (L;, 9j). The method is also configured to check if the grid is covered for the first time or the grid had already been covered by another location orientation pair (L;, 9j). A diagrammatic representation of the method is shown in FIG. 3 A. In one embodiment if the grid is covered for the first time by the location orientation pair (L;, 9j) the vote value is given by
[67] In another embodiment if the grid is covered by a previous location orientation
pair (L;, 9j) the vote value is given by
where H(k,i) is a grid covered by a previous location orientation pair (L;, 9j).
The vote value for each grid is read by the processing module and stored in a memory module. In step 203 the reading and storing process in step 202 is repeated for each camera and at all predetermined configurations. In step 204 an accumulated vote value is calculated by the processing module for each grid G^. The accumulated vote value for
each grid is obtained by adding the vote values provided for each grid. The accumulated vote value is given by
[68] In step 205 a vote value provided to the cameras set at a position and
orientation is read and stored in a memory module by the processing module. The vote value is based on the accumulated vote value for each grid and also on whether the

camera has already been covered previously by any grid. In various embodiments the method includes performing reverse voting by each grid for a location orientation pair (L;, 0j) that covers the grid. In this step the accumulated vote value of the grid obtained in step 204 is distributed among the cameras set at a position and orientation i.e. for each location orientation pair by the processing module. The method in step 205 is also configured to check if the location orientation pair is covered for the first time by the grid or is already been covered by another grid. In one embodiment if the location orientation pair is covered for the first time by the grid the vote value is given by
[69] In another embodiment if the location orientation pair is covered by a previous
grid the vote value is given by
where H^i) is a grid covered by a previous location orientation pair (L;, 9j). The generalized equation to calculate vote for a single location orientation pair is given by:
a and /? are the regularization parameters to model the camera coverage in a 2-D space. The parameter a controls total coverage and /? controls overlap ratio. In step 206 the process in step 207 is repeated iteratively for all the grids. An accumulated vote value for each of the location orientation pair is calculated in step 207. The accumulated vote value for each location orientation pair is calculated by adding all vote values from the associated grids given by

[70] In step 208 a location orientation pair (L;, 9j) that has the maximum
accumulated vote value is selected. The camera at location U is set with the corresponding orientation d\. In various embodiments the orientation of the camera at that location is set. The method iterates from step 202 to 208 for all the other cameras and for all predetermined orientations. The execution order of cameras is as shown in FIG. 3B. In various embodiments for each iteration the orientation of at least one camera is set. The iteration continues until all the cameras are set and the maximum coverage of the surveillance area is obtained. In various embodiments the cameras whose orientation is set is not considered for the next iteration. In various embodiments the method includes having a predetermined threshold value for the votes obtained by each grid. If the vote value of a grid is greater than the predetermined threshold value then the grid is grouped as covered grid.
[71] The method 200 may regulate the coverage gain and overlap ratio of the
surveillance system. Coverage overlap is the intersection of the field of view of the cameras. Overlap ratio is the ratio between the intersection of the field of view of at least two cameras and the total surveillance area covered by the at least two cameras. The minimization and the maximization arrangements of the coverage overlapping are regularized using (3 parameter. A high value for a increases the coverage gain and a high value for (3 reduces overlap ratio. A significantly high value for a, nullifies the impact of (3 on the total coverage. By adjusting a and (3 values, coverage at the surveillance area may be designed based on the application requirements. In some embodiments of the method the portion of the surveillance area that is covered by minimum number of (L;, 9j) pair gets high priority. The selection of coverage areas is determined by the regularizing parameters. The proportion of a and (3 determines the selection of coverage areas. In one embodiment the method functions in a minimizing overlap mode. In the minimizing overlap mode the value of a is set as zero, the method selects coverage areas in the order of minimizing overlap ratio and the coverage obtained is not maximum. In another embodiment the method functions in a maximizing coverage mode. In this mode the

value of /? is zero and the coverage gain is maximized. The regularization parameters may be set based on the surveillance application requirements.
[72] In various embodiments the surveillance area includes one or more critical
regions. Critical regions are high priority regions and require complete coverage. The threshold for setting the grids in the critical region as covered rCOv is 100%. The location orientation pair (Li, 6f) that covers the critical region is configured to receive a weighted vote from the grids in the critical region. Hence, the vote from the critical region may be high compared to the other regions. The reverse vote from the grids in critical region is
where ' Wc' is the weight factor for the vote from the grids in critical region.
[73] In some embodiments critical regions are covered by at least two cameras to
maximize depth monitoring of the critical regions. The method associates a coverage variable Gcov for each grid that is initialized with zero, ucov ' is incremented to one when the grid is set as covered. In some embodiments for a region to be covered by m cameras, the Gcov associated with the grids is initialized with 1-m. These grids require 'm' coverage increments to become one and to be considered as a covered grid. The advantage of the method is that it may regulate coverage gain and overlap ratio based on the application requirement. Also the method gives preference to cover the unique regions that has lesser chance to be covered. Further the order of execution is not in the raster scan order of camera location.
[74] In various embodiments a surveillance system 300 is disclosed. The
surveillance system as shown in FIG. 4 has a plurality of cameras 301-1, 301-2,...301-8 fixed at predetermined locations in a surveillance area 302 and connected over a network 303 to a computing system 304. The plurality of cameras 301-1, 301-2,...301-8 are configured to be oriented at a plurality of predetermined orientations. The computing system 304 that connects the plurality of cameras 301-1, 301-2,...301-8 over a network 303 includes a control module 305, a receiving module 306, a memory module 307 and a

processing module 308. The control module 305 is configured to set the plurality of cameras 301-1, 301-2,...301-8 to a first orientation and move each of the plurality of cameras 301-1, 301-2,...301-8 one at a time to a next orientation in the plurality of predetermined orientations of the cameras 301-1, 301-2,...301-8. The receiving module 306 is configured to read a coverage value of each of the plurality of cameras 301-1, 301-2,...301-8 set at a first orientation and next predetermined orientations. The memory module 307 is configured to store one or more values.
[75] In various embodiments the processing module 308 is configured to calculate a
total coverage value from the coverage values obtained at the first orientation of each of the plurality of cameras 301-1, 301-2,...301-8, store the total coverage value as a maximum coverage value in the memory module 307. The processing module 308 is then configured to move one of the plurality of cameras 301-1, 301-2,...301-8 to a next orientation in the plurality of predetermined orientations by initiating a command to the control module 305. In various embodiments with one of the cameras moved to the next orientation the processing module 308 reads the coverage value of each of the plurality of cameras 301-1, 301-2,...301-8 obtained from the receiving module 306 and calculate a total coverage value from the coverage values obtained from each of the plurality of cameras 301-1, 301-2,...301-8. The processing module 308 is then configured to compare the maximum coverage value and the calculated total coverage value. The larger of the compared total coverage values is stored as the maximum coverage value and an orientation corresponding to the stored total coverage value is selected for the one of the plurality of cameras 301-1, 301-2,...301-8. In various embodiments the processing module 308 initiates a command to the control module 305 until the orientations of all the cameras 301-1, 301-2,...301-8 are selected iteratively to obtain the maximum coverage of the surveillance area 302.
[76] In various embodiments the memory module 307 is configured to store the
maximum coverage value and the maximum of the compared coverage value. The processing module selects camera orientations that produce minimal overlapped coverage region at each iteration. In various embodiments the system is capable of conducting a

search for the locations that provide maximum coverage at an orientation to obtain the maximum coverage of the surveillance area 302.
[77] A surveillance system 300 as shown in FIG. 3 is disclosed. The system 300
includes a plurality of cameras 301-1, 301-2,...301-8 fixed at predetermined locations in a surveillance area 302 and configured to be oriented at a plurality of predetermined orientations. The plurality of cameras 301-1, 301-2,...301-8 are connected over a network 303 to a computing system 304. The computing system 304 includes a control module 305, a receiving module 306, a memory module 307 and a processing module 308. The control module 305 is configured to set the plurality of cameras 301-1, 301-2,...301-8 to a first orientation and move each of the plurality of cameras 301-1, 301-2,...301-8 one at a time to a next orientation in the plurality of predetermined orientations of the cameras 301-1, 301-2,...301-8. The receiving module 306 is configured to read a coverage value of each of the plurality of cameras 301-1, 301-2,...301-8 set at a first orientation and next predetermined orientations. The memory module 307 is configured to store one or more values.
[78] In various embodiments the processing module 308 is configured to divide the
surveillance area 302 into one or more grids. The processing module 308 is further configured to read and store in a memory module 307 a vote value that is provided to the one or more grids by a first camera and each of the plurality of cameras 301-1, 301-2,...302-8 at a first orientation and at all predefined orientations. The vote value is based on a percentage of coverage by the first camera and each of the plurality of cameras 301-1, 301-2,...302-8 at a first orientation and at all predefined orientations. In various embodiments the vote value is also based on whether the grid has already been covered by any camera 301-1, 301-2,...301-8 configured previously at an orientation. An accumulated vote value is calculated for each grid by adding the vote values for each grid. The processing module 308 further reads and stores in the memory module 307 a vote value provided to the cameras 301-1, 301-2,...301-8 that is set at a position and orientation by each of the one or more grids, based on the accumulated vote value for each grid. In various embodiments the vote value provided to the cameras is also based

on whether the camera 301-1, 301-2,...301-8 has already been covered previously by any grid.
[79] In various embodiments the processing module is configured to calculate an
accumulated vote value for each of the plurality of cameras 301-1, 301-2,...301-8 for all defined orientations by adding the vote values for each camera. In various embodiments an orientation is selected for one of the plurality of cameras 301-1, 301-2,...301-8 corresponding to the accumulated vote value that is maximum. Further the processing module 308 initiates a command to the control module 305 until the orientations of the plurality of cameras 301-1, 301-2,...301-8 are selected to obtain the maximum coverage of the surveillance area 302. In various embodiment the memory module 307 is configured to store the vote value provided to the one or more grids, the vote value provided to the cameras 301-1, 301-2,...301-8 set at a position and orientation and the accumulated vote value.
[80] In various embodiment the processing module 308 is configured to provide
vote values to the grids within the one or more grids. In one embodiment if the grid has not been covered previously by any camera, set at an orientation the vote value to the grid is given by equation (2). In another embodiment if the grid has already been covered previously by any camera, set at an orientation the vote value to the grid is given by equation (3). In various embodiments of the system the processing module 308 is configured to provide vote value to the cameras set at a position and orientation given by equation (7). In various embodiments the system is configured to cover one or more critical regions by 2 or more cameras 301-1, 301-2,. ..301-8.
EXAMPLES
[81] Example 1: Modeling the coverage area of a visual surveillance system and comparing the coverage results in multiple optimization techniques
[82] Camera coverage was modeled deterministically using an enhanced pinhole
camera model. A triangular coverage was considered to reduce computational

complexity. FIG. 5 A shows triangular coverage area of a sensor placed at origin o with range r and horizontal angle of the field of view a. The value of a may be calculated as:
where Wj is width of the image sensor and / is focal length of the camera. In FIG. 5A, the coordinate positions of A and B were considered as (r, y) and (r, -y). The value of y was evaluated as:
FIG. 5B is an example for coverage without overlap and with overlap between two visual sensors positioned at oxand o2. The degree of coverage overlapping relies on the sensor location, type and configuration. The visual sensors are considered to be of same type and had range r. When the sensors were positioned at least 2 xr distance apart the coverage overlap did not occur. Another overlap constraint is camera orientation. Albeit the cameras positions are within the range distance 2 x r, the coverage overlap occurred when the coverage angle fell within the following range:
For camera at position 0X

[83] To conduct experimental study a 300 x 300 binary image was considered as the surveillance room. Cameras of same model were used for coverage and its locations were preset. 16 cameras with field of view (a) 50° and range 100 px, were placed at equal distance. The coverage area was approximated to triangular shape. The coordinates of camera locations in raster scan order are (1,1), (1,100), (1,200), (1,300), (100,1), (100,100), (100,200), (100,300), (200,1), (200,100), (200,200), (200,300), (300,1), (300,100), (300,200), (300,300). Each camera may take finite orientations. For conducting experimental study only eight camera orientations were considered. The orientations are: (0°, 45°, 90°, 135°, 180°, 225° , 270° , 315°).
[84] The traditional local and global greedy techniques start optimization from the camera at first location and the process continues sequentially in the raster scan order. Since the local greedy technique is not considering the overlap with the previous coverage, the total coverage gain obtained was not optimal. The global greedy technique selects the orientation which provides maximum non overlapped area. Hence, global greedy method produced better coverage gain and less overlap ratio compared to local greedy technique.
[85] The proposed method is named as Alternate Global Greedy (AGG) method. The AGG method produced better coverage result with minimum overlap compared to global greedy and local greedy methods. In the greedy methods the coverage results rely on the order of camera selected for optimization. In traditional greedy methods, a camera orientation is selected for a location based on the total coverage gain. Whereas in AGG method, location is selected for a camera orientation in such a way that it maximizes the coverage.
[86] The proposed grid-based method is named as Global Grid Voting (GGV) method. For the experimental study, the surveillance region was divided into 900 grids of size 10x10. In GGV method, each camera location and orientation is considered as (Li,

6\) pair. The first step of GGV algorithm is identification of (Li, 6\) pair. In the experimental study, 16 cameras and 8 camera orientations were considered. Hence 128 (Li, 6]) pairs were generated. The values of regularization parameter were first set as a = 0.9 and /? = 0.1. Then, the proposed algorithm is operated in the maximizing coverage mode (GGV-max) and attained maximum coverage gain. The coverage result achieved by GGV-max mode was higher than the other existing greedy methods. When the values of regularization parameters were set as a = 0.1 and /? = 0.9, the algorithm produced minimum overlap ratio, which is the minimizing overlap mode (GGV-min). The comparison of coverage results obtained for each optimization algorithm is listed in
m 11 -i
[87] FIG. 6A-6F shows the coverage results obtained with different optimization methods. The camera locations are marked in circles and covered areas were displayed using triangles. FIG. 6E and 6F show the coverage results of proposed GGV-max mode and GGV-min mode. Unlike existing greedy methods, GGV method does not set camera

orientation in the raster scan order of camera location. The execution order and selected orientation of different optimization results are listed in Table 2. The coverage outputs of proposed GGV-max and GGV-min mode at each iteration are shown in FIG. 7A and 7B respectively. In GGV-max mode, the camera orientation that is covering unique regions gets highest priority. Whereas in GGV-min mode, the camera orientation that reduces the coverage overlap gets the highest priority.

[88] FIG. 8A shows the plot of coverage area obtained with different methods by adding each camera to the surveillance system per iteration. Similarly, FIG. 8B shows the plot for comparing overlapped area and overlapping ratio with different optimization techniques respectively. In FIG. 8A, the GGV-min algorithm produced maximum coverage gain until seventh iteration. Thereafter, the coverage result decreases gradually compared to the other optimization algorithm. The GGV-min algorithm only selects orientations which do not overlap with previous coverage. However, such orientations do not guarantee maximum coverage results. The GGV-max mode generated maximum coverage results and the proposed AGG method produced second maximum coverage results. In FIG. 8B the GGV-min algorithm produced minimum overlap with the previous cameras coverage. In GGV-min algorithm, overlapping with the previous cameras coverage is zero until twelfth iteration. The remaining section explains the behavior of GGV method for varying grid size and regularization parameters.
[89] i. Selection of grid size:
[90] Since GGV method is a grid based method, the selection of grid size has an impact over the coverage results. In the experimental study, the GGV method was tested using 5 different grid sizes (30 x 30, 20 x 20, 10 x 10, 5 x 5, 1 x 1). When the grid size was lxl, it indicates that each pixel is considered as a grid and the total number of grids were 90000. Then, maximizing grid coverage becomes maximizing pixel coverage. To experiment the efficiency of proposed GGV method with varying grid sizes, the values of regularization parameters were initially set as a = 0.9 and /? = 0.1. The algorithm operated in GGV-max mode when the a value was high. The coverage results of

proposed method for different grid size are listed in Table. 3. Maximum coverage results were obtained with a moderate grid size. The maximum coverage result was obtained with grid size 20 x 20 and 10 x 10 .
[91] Subsequently, the values of regularization parameters were set to a = 0.1 and /? = 0.9 and experimented the algorithm in GGV-min mode. The coverage results of the algorithm with aforementioned regularization parameter values are listed in Table. 4. The algorithm produced minimum overlap ratio with grid size 10x10 and 5x5. The overlap ratio significantly reduced from 11.59% to 9.17% when the grid size changed from 30 x 30 to 10 x 10. Based on the experiments conducted, we conclude that the grid size should not be too small or large to attain maximum coverage results. The algorithm produced better coverage results with grid size 10 x 10. Hence, for the experimental study the grid size was fixed as 10 x 10.
Table. 4: Results for Grid Size Selection to Minimize Coverage Overlap in GGV-Min Method

[92] n. Selection of regulanzation parameters: The coverage and overlap ratio of the proposed GGV algorithm is determined by the regularization parameters. The total coverage was at maximum for high 'a' and low '/?' value. The overlap ratio was at minimum for low 'a' and high '/?' value. The heatmap for the total coverage gain attained by the proposed algorithm with varying 'a' and '/? values are shown in FIG. 9A. The range of 'a' and '/?' was set between 0 and 1. The total coverage was high for high 'a' values. A significantly high value for a nullifies the impact of /? on the total coverage. In the experimental study, the maximum coverage obtained was 64437 pixels. FIG. 9B shows the heatmap of total overlapped area. The minimum overlapped area obtained was 4 pixels when the value of 'a = 0'. But in the scenario, the total coverage gain obtained was minimum. The overlapped area and overlap ratio was minimum for high '/?' values. Hence, 'a' and '/?' values may be set based on the coverage and overlap requirements of the application.
[93] iii. Coverage at unique regions: The proposed GGV algorithm gives first preference to the unique regions which has lesser chance to be covered. During the first iteration and the subsequent iterations of the forward voting process each grid may be covered by different number of cameras. For example in a 10X10 grid in the first iteration grid (1,1) is covered by 4, grid (1,7) is covered by 4, grid (3,1) is covered by 3, grid (5,1) is covered by 4, grid (5,1) is covered by 1 etc. Each grid may be covered by different number of camera. In a surveillance area with 8 cameras there may be grids that may be covered by all 8 cameras, some may be covered by 4 and there are few grids that

are covered by only one camera. The number of cameras that may cover a grid is updated by the GGV- algorithm after each iteration. The region which may be covered by minimum number of cameras i.e. 1 or 2 cameras is the unique region. Those are the regions that may be covered first. During the reverse voting process the grids that are covered by minimum number of cameras say 1 or 2 are covered during first few iterations. Heatmaps were generated for the number of cameras that covered the grids during each iteration and also the grid that are covered during each iteration. Table. 5 is a table form of the heatmap for the first three iteration. Table. 6 is a table form of the heatmap for the grids that are covered during the first three equations. Comparing Table 5 and Table 6 the unique regions that are covered by only one or 2 cameras is covered during the first iteration and the next iterations. It is clear from Table. 5 and Table. 6 that the algorithm gives preference to the unique regions that may be covered by minimum number of cameras.

[95] In this section, we discuss the empirical coverage results obtained for different optimization methods on an examination hall. The dimension of examination hall was 1070x970 centimeter and the hall was under the surveillance of eight static cameras. FIG. 10A shows the layout of the examination hall. The seating arrangements of students are displayed using rectangles.
[96] The 'AXIS P1357' static network cameras were used for the study. The image sensor of the camera is 1/3.2" RGB CMOS and the lens is varifocal (2.8- 8 mm) with

horizontal field of view ranges between 92° to 32°. The working range of camera was fixed to 634px and horizontal field of view was set to 54° using Eq. 6 and Eq. 7. Each camera took 72 orientations ranges from 0° to 355° with an orientation step size 5°.
[97] Different coverage optimization algorithms were tested against the examination hall and the result is shown in FIG. 10B-10F. The coverage results obtained with optimization algorithms is listed in Table 7. The LG and GG method covered only 70.3% and 82.55% area of the examination hall. Compared to the local greedy method, global greedy method produced better coverage gain and lower overlapping ratio. The proposed AGG algorithm covered 90.11% area of the examination hall. The proposed alternate global greedy method produced further improved result with a comparatively high coverage gain and low overlapping ratio.
[98] The grid size for the experimental study was set to 10x10. Since, in the previous experimental study the GGV algorithm produced maximum coverage result when the

grid size was 10x10. The GGV algorithm operated in GGV-max mode produced maximum coverage gain. It covered 95.24% area of the examination hall. The minimum overlap ratio was achieved when the algorithm operates in GGV-min mode. In GGV-min mode the overlap ratio was 0.12% and the total area covered was 65.53%. By adjusting the values of regularization parameter the overlap ratio may be varied from 0.12% -49.1% and coverage gain may be from 65.53% - 95.24%. FIG. 10E shows the maximum covered area achieved in the examination hall with the GGV algorithm, when the regularization parameters were set as a = 0.9 and /? = 0.1. In examination hall surveillance applications, the prime importance is to cover maximum area. Therefore, to maximize coverage high value was given to a and a comparatively low or zero value was assigned to /?.
[99] i. Coverage at critical regions:
[100] The student occupied regions in the examination hall were set as critical regions. The orientation of cameras in the examination hall were aligned in such a way that it maximizes camera coverage at critical region. Therefore, the GGV-max algorithm uses weighted vote as in Eq. 18 for the reverse voting process. FIG. 11A shows the coverage result of GGV-max algorithm in the examination hall without setting critical regions. Whereas, the coverage result after setting the high priority critical regions is shown in FIG. 11B. After setting priorities to the critical regions, the camera orientations were more aligned towards the critical regions. The comparison of coverage at the critical regions before and after setting critical regions is shown in FIG. 11C. The GGV-max algorithm improved the coverage at critical region from 94.6% to 98.4% after setting the priority. Table 8 shows the number of cameras that cover a grid at the critical region. After setting priority to the critical regions, the GGV-max algorithm reduced the number of uncovered grids in critical region from 124 to 51.

[101] Although the detailed description contains many specifics, these should not be construed as limiting the scope of the invention but merely as illustrating different examples and aspects of the invention. It should be appreciated that the scope of the invention includes other embodiments not discussed herein. Various other modifications, changes and variations which will be apparent, to those skilled in the art, may be made in the arrangement, operation and details of the system and method of the present invention disclosed herein without departing from the spirit and scope of the invention as described here. While the invention has been disclosed with reference to certain embodiments, it will be understood by those skilled in the art, that various changes may be made and equivalents may be substituted, without departing from the scope of the invention. In addition, many modifications may be made to adapt to a particular situation or material the teachings of the invention without departing from its scope.

1. A method of identifying optimal camera configurations for maximum coverage of a surveillance area, the surveillance area having a plurality of cameras fixed at predetermined locations, each of the plurality of cameras capable of being oriented at a plurality of predetermined orientations, the method comprising:
a. reading a coverage value of each of the plurality of cameras set at a first
orientation (101), by a receiving module of a computing system, the coverage
value of each of the plurality of cameras indicating an area of coverage of the
surveillance area;
b. calculating a total coverage value (102) from the coverage values obtained at
the first orientation of each of the plurality of cameras by a processing module of
the computing system;
c. storing the total coverage value as a maximum coverage value (103) in a
memory module of the computing system;
d. moving one of the plurality of cameras to a next orientation (106) in the
plurality of predetermined orientations of the cameras, by a control module of the
computing system;
e. reading the coverage value (106) of each of the plurality of cameras, by the
receiving module of the computing system;
f calculating a total coverage value (107) from the coverage value obtained from
each of the plurality of cameras by the processing module of the computing
system;
g. comparing the maximum coverage value and the calculated total coverage
value (108) by a processing module of the computing system, and storing the
larger of the compared total coverage values as the maximum coverage value
(109) in the memory module of the computing system;
h. selecting an orientation corresponding to the stored total coverage value (110),
for the one of the plurality of cameras; and
i. repeating steps d to h iteratively for each of the plurality of cameras at the next
orientation (111) and through the plurality of predetermined orientations (112);
and

where CL coverage gain, O is the orientation and L is the location of cameras.
3. A method of identifying optimal camera configurations (200) for maximum covera^ of a surveillance area monitored using a surveillance system having a plurality ( cameras fixed at predetermined locations and orientations, a computing systei comprising a receiving module, a processing module, a control module and a memoi module, the method comprising the steps of:
a. dividing the surveillance area into one or more grids (201) by the processiri
module;
b. reading and storing in a memory module a vote value provided to the one (
more grids (202) by the processing module, based on a percentage of coverage b
a first camera at a first orientation, and also on whether the grid has already bee
covered by any camera configured previously at an orientation;
c. repeating step b for the plurality of cameras (203) fixed at predetermine
locations and all predefined orientations of the plurality of cameras;
d. calculating an accumulated vote value for each grid (204) by adding the voi
values for each grid obtained from steps b - c;
e. reading and storing in a memory module a vote value provided to the camen
(205) set at a position and orientation by the processing module, based on tr
accumulated vote value for each grid and also on whether the camera has alread
been covered previously by any grid;
f repeating step e iteratively for the one or more grids (206); h. calculating an accumulated vote value for each of the plurality of cameras ft all defined orientations (207) by adding the vote values for each cameras obtaine from steps e and f; and

i. selecting an orientation (208) for one of the plurality of cameras corresponding
to the accumulated vote value that is maximum;
j. repeating steps b to i iteratively (209) through all defined orientations for each of the
plurality of cameras; and
k. obtaining maximum coverage of the surveillance area (210).
4. The method of claim 3, wherein the vote value provided to the grids within the one or
more grids by the processing module is
Vi,j{p(k,i)) = Q,i(^i' 0/), if the grid has not been covered previously by any camera, set at an orientation, wherein Ckii(Lt, 6j) is the percentage of grid G^^ covered by the camera;
vij(H(k,i-)) = -Ck,i{Li'ei) > if the §rid has already been covered previously by any camera, set at an orientation, wherein H^k^ is a grid that has already been covered previously by any camera, set at an orientation.
5. The method of claim 3, wherein the vote value provided to the cameras set at a
position and orientation by the processing module is
parameters where a controls a total coverage and /? controls an overlap ratio.
6. The method of claim 5, wherein the values of a and /? ranges between 0 and 1.
7. The method of claim 6, wherein the value of a is in a range 0.5 to 1 the method operates in a maximizing coverage mode, wherein the cameras are set at orientations that cover the regions that have a lesser chance of being covered.

8. The method of claim 6, wherein the value of /? is in a range 0.5 to 1 and a is in a range 0 to 0.4 the method operates in a minimizing overlap mode, wherein the cameras are set at orientations that reduce the coverage overlap.
9. The method of claim 3, wherein the surveillance area comprises one or more critical regions.

10. The method of claim 9, wherein the minimum coverage threshold Tcov of the one or more critical regions is 100%.
11. The method of claim 9, wherein performing voting for a camera set at an orientation at the critical region by the grids is weighted
where Wc is the weight factor for the vote from the grids.
12. The method of claim 9, wherein the critical region is covered by 2 or more cameras in the network.
13. The method of claim 9, wherein covering the critical region by 'm' cameras, the coverage value associated with each grid is initialized with ' 1-m'.
where Gcov is the coverage value associated with each grid.
14. A surveillance system comprising
a plurality of cameras (301-1, 301-2,...301-8) fixed at predetermined locations in a surveillance area (302) and configured to be oriented at a plurality of predetermined orientations;
a computing system (304) connecting the plurality of cameras (301-1, 301-2,.. .301-8) over a network (303) and comprising

a control module (305) configured to set the plurality of cameras (301-1, 301-2,...301-8) to a first orientation and move each of the plurality of cameras (301-1, 301-2,...301-8) one at a time to a next orientation in the plurality of predetermined orientations of the cameras (301-1, 301-2,...301-8); a receiving module (306) configured to
read a coverage value of each of the plurality of cameras (301-1, 301-2,...301-8) set at a first orientation and next predetermined orientations; a memory module (307) configured to store one or more values; a processing module (308) configured to:
calculating a total coverage value from the coverage values
obtained at the first orientation of each of the plurality of cameras
(301-1, 301-2,...301-8);
store the total coverage value as a maximum coverage value in the
memory module (307);
move one of the plurality of cameras (301-1, 301-2,...301-8) to a
next orientation in the plurality of predetermined orientations by
initiating a command to the control module (305);
read the coverage value of each of the plurality of cameras (301-1,
301-2,.. .301-8) obtained from the receiving module (306);
calculate a total coverage value from the coverage value obtained
from each of the plurality of cameras (301-1, 301-2,... 301-8);
compare the maximum coverage value and the calculated total
coverage value and store the larger of the compared total coverage
values as the maximum coverage value; and
select an orientation corresponding to the stored total coverage
value, for the one of the plurality of cameras (301-1, 301-2,... 301-
8), wherein

the processing module (308) initiates a command to the control module (305) until the orientations of the plurality of cameras (301-1, 301-2,... 301-8) are selected to obtain the maximum coverage of the surveillance area (302).
15. The system of claim 14, wherein the memory module (307) is configured to store the maximum coverage value.
16. A surveillance system comprising
a plurality of cameras (301-1, 301-2,...301-8) fixed at predetermined locations in a surveillance area (302) and configured to be oriented at a plurality of predetermined orientations;
a computing system (304) connecting the plurality of cameras (301-1, 301-2,.. .301-8) over a network (303) and comprising
a control module (305) configured to set the plurality of cameras (301-1, 301-2,...301-8) to a first orientation and move each of the plurality of cameras (301-1, 301-2,...301-8) one at a time to a next orientation in the plurality of predetermined orientations of the cameras (301-1, 301-2,...301-8); a receiving module (306) configured to
read a coverage value of each of the plurality of cameras (301-1, 301-2,...301-8) set at a first orientation and next predetermined orientations; a memory module (307) configured to store one or more values; a processing module (308) configured to:
divide the surveillance area (302) into one or more grids; read and store in a memory module (307) a vote value provided to the one or more grids based on a percentage of coverage by a first camera and each of the plurality of cameras (301-1, 301-2,... 301-8) at a first orientation and at all predefined orientations, and also on whether the grid has already been covered by any camera (301-1, 301-2,.. .301-8) configured previously at an orientation;

calculate an accumulated vote value for each grid by adding the vote values for each grid;
read and store in the memory module (307) a vote value provided to the cameras (301-1, 301-2,...301-8) set at a position and orientation by each of the one or more grids, based on the accumulated vote value for each grid and also on whether the camera (301-1, 301-2,...301-8) has already been covered previously by any grid;
calculate an accumulated vote value for each of the plurality of cameras (301-1, 301-2,...301-8) for all defined orientations by adding the vote values for each camera; and select an orientation for one of the plurality of cameras (301-1, 301-2,...301-8) corresponding to the accumulated vote value that is maximum, wherein the processing module (308) initiates a command to the control module (305) until the orientations of the plurality of cameras (301-1, 301-2,... 301-8) are selected to obtain the maximum coverage of the surveillance area (302).
17. The system of claim 16, wherein the memory module (307) is configured to store the vote value provided to the one or more grids, the vote value provided to the cameras (301-1, 301-2,.. .301-8) set at a position and orientation and the accumulated vote value.
18. The system of claim 16, wherein the processing module (308) is configured to provide vote values to the grids within the one or more grids is given by
Vi,j{p(k,i)) = Ck,i{Li> Qj), if the grid has not been covered previously by any camera, set at an orientation, wherein Ckii(Lt, 6j) is the percentage of grid G^^ covered by the
camera;
vij(H(k,i-)) = -Ck,i{Li'ei) > if the §rid has already been covered previously by any camera, set at an orientation,

wherein H^k^ is a grid that has already been covered previously by any camera, set at an orientation.
19. The system of claim 16, wherein the processing module (308) is configured to provide vote value to the cameras set at a position and orientation given by
parameters where a controls a total coverage and /? controls an overlap ratio.
20. The system of claim 16, wherein the system is configured to cover one or more critical regions by 2 or more cameras (301-1, 301-2,...301-8).