Detecting potential changed objects in images
A method of detecting potential changed objects in a first image and a second image, where the first and second images at least partially overlap. The method includes obtaining data describing first and second images and detecting sets of objects in images. A common coordinate system for the sets of objects is calculated. Objects in the sets are eliminated based on positions of the objects and objects outside an area of overlap of the images are eliminated. Data indicating the zero or more remaining objects in the sets is output to indicate potential changed objects.
1. A method of detecting potential changed objects in a first image and a second image, where the first and second images at least partially overlap, the method including:
obtaining data describing a first image and data describing a second image;
detecting a first set of objects in the first image data;
detecting a second set of objects in the second image data;
calculating a common coordinate system for the first set of objects and the second set of objects;
converting locations of the objects in the first set and/or the second set into the common coordinate system;
targeting of said objects in the first set and the second set based on positions of the objects in the sets;
targeting of said objects in the first set and the second set that are located outside an area of overlap of the first image and the second image; and
outputting data indicating zero or more of said objects in the first set and the second set as potential changed objects,
wherein the step of eliminating zero or more objects in the first set and the second set based on positions of the objects in the sets includes:
calculating a distance from at least one object in the first set and a nearest neighbour of that object in the second set;
calculating a distance from at least one object in the second set and a nearest neighbour of that object in the first set;
eliminating a said object in the first set if the distance between that object and the nearest neighbour of the object meets one or more culling criteria; and
eliminating a said object in the second set if the distance between that object and the nearest neighbour of the object meets the culling criteria.
2. A method according to claim 1 , wherein the data output includes all the objects not eliminated from the first set and all the objects not eliminated from the second set.
3. A method according to claim 1 , wherein the distance between the objects meets the culling criteria if it is below a threshold value.
4. A method according to claim 3 , wherein the threshold value corresponds to a registration error value of a registration process used to calculate the common coordinate system.
5. A method according to claim 4 , wherein the threshold value corresponds to 10 pixels in the first and second image data.
6. A method according to claim 1 , wherein the detecting of the first set of objects in the first image data and/or the detecting of the second set of objects in the second image data is performed using a Triple Window Filter-based object detection algorithm.
7. A method according to claim 1 , wherein the calculating of the common coordinate system includes generating a homography matrix.
8. A method according to claim 1 , including allowing a user to select an object detection technique from a set of object detection techniques for detecting the first set and/or the second set of objects.
9. A method according to claim 1 , including allowing a user to select from a set of registration techniques to determine the calculation of the common coordinate system.
10. A method according to claim 1 , comprising:
executing the step of eliminating zero or more said objects in the first set and the second set based on positions of the objects in the sets in a sequential manner.
11. A method according to claim 1 , comprising:
executing the step of eliminating zero or more said objects in the first set and the second set based on positions of the objects in the sets in a parallel manner.
12. A system configured to detect potential changed objects in a first image and a second image, where the first and second images at least partially overlap, the apparatus including:
a first device configured to obtain data describing a first image and data describing a second image;
a second device configured to detect a first set of objects in the first image data;
a third device configured to detect a second set of objects in the second image data;
a fourth device configured to calculate a common coordinate system for the first set of objects and the second set of objects;
a fifth device configured to convert locations of the objects in the first set and/or the second set into the common coordinate system;
a sixth device configured to target of said objects in the first set and the second set based on positions of the objects in the sets;
a seventh device configured to target of said objects in the first set and the second set that are located outside an area of overlap of the first image and the second image; and
an eighth device configured to output data indicating zero or more of said objects in the first set and the second set as potential changed objects,
wherein at least one of the sixth, seventh, and/or eighth devices is configured to eliminate zero or more objects in the first set and the second set based on positions of the objects in the sets and is arranged:
to calculate a distance from at least one object in the first set and a nearest neighbour of that object in the second set;
to calculate a distance from at least one object in the second set and a nearest neighbour of that object in the first set;
to eliminate a said obiect in the first set if the distance between that object and the nearest neighbour of the object meets one or more culling criteria; and
to eliminate a said object in the second set if the distance between that object and the nearest neighbour of the object meets the culling criteria.
13. An apparatus according to claim 12 , wherein the data output includes all the objects not eliminated from the first set and all the objects not eliminated from the second set.
14. An apparatus according to claim 12 , wherein the detecting of the first set of objects in the first image data and/or the detecting of the second set of objects in the second image data is performed using a Triple Window Filter-based object detection algorithm.
15. An apparatus according to claim 12 , wherein the calculating of the common coordinate system includes generating a homography matrix.
16. An apparatus according to claim 12 , including allowing a user to select an object detection technique from a set of object detection techniques for detecting the first set and/or the second set of objects.
17. An apparatus according to claim 12 , including allowing a user to select from a set of registration techniques to determine the calculation of the common coordinate system.
18. A non-transitory computer readable memory causing a computer to execute a procedure for detecting potential changed objects in a first image and a second image, where the first and second images at least partially overlap, the procedure comprising:
obtaining data describing a first image and data describing a second image;
detecting a first set of objects in the first image data;
detecting a second set of objects in the second image data;
calculating a common coordinate system for the first set of objects and the second set of objects;
converting locations of the objects in the first set and/or the second set into the common coordinate system;
targeting of said objects in the first set and the second set based on positions of the objects in the sets;
targeting of said objects in the first set and the second set that are located outside an area of overlap of the first image and the second image; and
outputting data indicating zero or more of said objects in the first set and the second set as potential changed objects,
wherein the step of eliminating zero or more said objects in the first set and the second set based on positions of the objects in the sets includes:
calculating a distance from at least one object in the first set and a nearest neighbour of that object in the second set;
calculating a distance from at least one object in the second set and a nearest neighbour of that object in the first set;
eliminating a said object in the first set if the distance between that object and the nearest neighbour of the object meets one or more culling criteria; and
eliminating a said object in the second set if the distance between that object and the nearest neighbour of the object meets the culling criteria.
19. A non-transitory computer readable memory according to claim 18 , wherein the data output includes all the objects not eliminated from the first set and all the objects not eliminated from the second set.
20. A non-transitory computer readable memory according to claim 18 , wherein the detecting of the first set of objects in the first image data and/or the detecting of the second set of objects in the second image data is performed using a Triple Window Filter-based object detection algorithm.