Camera image georeferencing systems
First Claim
1. A method of deriving at least one camera model, for security surveillance in at least one underlying terrain and georeferencing pixel coordinates in a camera view of a single security surveillance camera, the camera view being deficient of located landmarks in sufficient number or desired dispersion to produce the at least one camera model, said method comprising the steps of:
- a) obtaining, from the single security surveillance camera, a principal camera image comprising a principal view of the at least one underlying terrain, the principal view comprising at least two located landmarks;
b) for each one of the at least two located landmarks of the principal view, producing, by at least one computer processor, a principal node associating physical coordinates of the located landmark in the at least one underlying terrain to pixel coordinates in the principal camera image;
c) generating, by at least one computer processor, a principal nodelist comprising all available principal nodes of the principal view;
d) obtaining from the single security surveillance camera, at least one auxiliary camera image comprising at least one auxiliary view of the at least one underlying terrain, the at least one auxiliary view comprising a plurality of located landmarks;
e) for each one of the plurality of located landmarks of the at least one auxiliary view, producing, by at least one computer processor, an auxiliary node associating physical coordinates of the located landmark in the at least one underlying terrain to pixel coordinates in the at least one auxiliary camera image;
f) generating, by at least one computer processor, at least one auxiliary nodelist comprising all available auxiliary nodes of the at least one auxiliary view;
g) deriving, by at least one computer processor, from both the principal nodelist and the at least one auxiliary nodelist, at least two alignment nodes, each one of the at least two alignment nodes exhibiting respectively the same pixel coordinates in the at least one auxiliary camera image as the pixel coordinates in the principal camera image;
h) geometrically aligning, by at least one computer processor, the principal view to the at least one auxiliary view through pairing of the at least two alignment nodes of the principal nodelist and the at least one auxiliary nodelist;
i) generating, by at least one computer processor, an augmented principal nodelist by adding the nodes of the at least one auxiliary nodelist to the principal nodelist using a first set of geometrical computations, wherein the augmented principal nodelist comprises at least one enlarged set of located landmarks associated with the principal view; and
j) using the augmented principal nodelist, obtaining, by at least one computer processor, the at least one camera model by at least one second set of geometrical computations;
k) wherein the at least one camera model is utilized in security surveillance in the at least one underlying terrain.
2 Assignments
0 Petitions
Accused Products
Abstract
A method for constructing a georeferencing-enabled camera model and its deployment in georeferencing targets located in video sequences due to a single video camera. A video surveillance camera is modeled by a collection of rays converging at a virtual camera point and the retina resolution cell coordinates associated with those rays wherein the ray equations are first established, in the course of a calibration process, for a given camera view, with the aid of other views of the same video surveillance camera, as necessary, and using such a model for mapping image coordinates to terrain coordinates and vice versa in the intended view or its adaptation for use in other views of the same video surveillance camera.
44 Citations
20 Claims
-
1. A method of deriving at least one camera model, for security surveillance in at least one underlying terrain and georeferencing pixel coordinates in a camera view of a single security surveillance camera, the camera view being deficient of located landmarks in sufficient number or desired dispersion to produce the at least one camera model, said method comprising the steps of:
-
a) obtaining, from the single security surveillance camera, a principal camera image comprising a principal view of the at least one underlying terrain, the principal view comprising at least two located landmarks; b) for each one of the at least two located landmarks of the principal view, producing, by at least one computer processor, a principal node associating physical coordinates of the located landmark in the at least one underlying terrain to pixel coordinates in the principal camera image; c) generating, by at least one computer processor, a principal nodelist comprising all available principal nodes of the principal view; d) obtaining from the single security surveillance camera, at least one auxiliary camera image comprising at least one auxiliary view of the at least one underlying terrain, the at least one auxiliary view comprising a plurality of located landmarks; e) for each one of the plurality of located landmarks of the at least one auxiliary view, producing, by at least one computer processor, an auxiliary node associating physical coordinates of the located landmark in the at least one underlying terrain to pixel coordinates in the at least one auxiliary camera image; f) generating, by at least one computer processor, at least one auxiliary nodelist comprising all available auxiliary nodes of the at least one auxiliary view; g) deriving, by at least one computer processor, from both the principal nodelist and the at least one auxiliary nodelist, at least two alignment nodes, each one of the at least two alignment nodes exhibiting respectively the same pixel coordinates in the at least one auxiliary camera image as the pixel coordinates in the principal camera image; h) geometrically aligning, by at least one computer processor, the principal view to the at least one auxiliary view through pairing of the at least two alignment nodes of the principal nodelist and the at least one auxiliary nodelist; i) generating, by at least one computer processor, an augmented principal nodelist by adding the nodes of the at least one auxiliary nodelist to the principal nodelist using a first set of geometrical computations, wherein the augmented principal nodelist comprises at least one enlarged set of located landmarks associated with the principal view; and j) using the augmented principal nodelist, obtaining, by at least one computer processor, the at least one camera model by at least one second set of geometrical computations; k) wherein the at least one camera model is utilized in security surveillance in the at least one underlying terrain. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16)
-
-
17. A method of georeferencing at least one target, of a camera view of a security surveillance camera, the camera view comprising an associated georeferencing-enabled camera model, the camera model relating image pixel-coordinates of the camera view to rays in three-dimensional space, relative to a virtual camera point, said method comprising the steps of:
-
a) identifying, by at least one computer processor, in camera space, a target blob comprising a discrete region of camera space associated with the at least one target of the camera view; b) determining, by at least one computer processor, for the target blob, at least one blob attribute in camera space, the at least one blob attribute at least-comprising pixel coordinates of a lowest blob point and pixel coordinates of a blob centroid point; c) mapping, by at least one computer processor, the pixel coordinates of the lowest blob point to a terrain point using the georeferencing-enabled camera model; d) computing, by at least one computer processor, a range value, R_low, comprising a distance from the terrain point to the virtual camera point; e) associating, by at least one computer processor, the same range value R_low to the blob centroid point; f) intersecting, by at least one computer processor, a ray extending through the pixel coordinates of the blob centroid point with a sphere centered on the virtual camera point and comprising a radius equal to R_low; and g) disambiguating, by at least one computer processor, the result by selecting from at least one resulting intersection of the ray, a ray-end point that exhibits the lowest Z coordinate value, wherein coordinates of the ray-end point comprise a location of the at least one target in object space. - View Dependent Claims (18)
-
-
19. A method of arriving at an approximation of a real size of a target, of a camera view of a security surveillance camera, the camera view comprising an associated camera model, the camera model relating image pixel-coordinates of the camera view to rays in three-dimensional space, relative to a virtual camera point, said method comprising the steps of:
-
a) identifying, by at least one computer processor, in camera space, at least one target blob comprising a discrete region of camera space associated with the at least one target of the camera view; b) determining, by at least one computer processor, for the at least one target blob, at least one blob attribute in camera space, the at least one blob attribute at least comprising pixel coordinates of a lowest blob point and pixel coordinates of a target bounding box comprising at least four corners; c) mapping, by at least one computer processor, the pixel coordinates of the lowest blob point to a terrain point using the associated camera model; d) computing, by at least one computer processor, a range value, R_low, comprising a distance from the terrain point to the virtual camera point; e) assigning, by at least one computer processor, the range value R_low to the at least four corners of the target bounding box; f) generating, by at least one computer processor, a spherical quadrilateral by intersecting rays associated with the at least four corners of the target bounding box by a sphere of radius R_low and centered on the virtual camera point, said spherical quadrilateral comprising a spherical quadrilateral area; g) disambiguating, by at least one computer processor, intersections of the sphere and the rays in above said step (f), by selecting from resulting intersections of the rays, the intersections exhibiting lower Z coordinate value; h) determining, by at least one computer processor, in camera space, i) a number of pixels in the target bounding box, and ii) a number of pixels spanned by the at least one target blob; and i) computing for object space, a mean frustum area associated with each pixel, by at least one computer processor, wherein said computing comprises dividing of the spherical quadrilateral area by the number of pixels in the target bounding box and multiplying the number of pixels claimed by the at least one target blob in camera space by an equivalent mean frustum area associated with each pixel.
-
-
20. A method, of deriving at least one georeferencing-enabled camera model, usable for security surveillance in at least one underlying terrain and georeferencing image pixel coordinates in a camera view from a virtual camera point of a single security surveillance camera, wherein the at least one georeferencing-enabled camera model comprises at least one lookup table linking the image pixel coordinates of a camera image to rays in three-dimensional space that in turn lead to corresponding points on an underlying terrain, said method comprising the steps of:
-
a) obtaining, by at least one computer processor, from the single security surveillance camera, a selected camera view of the underlying terrain, the selected camera view comprising an associated camera image; b) recording, by at least one computer processor, zoom settings of the single security surveillance camera, while retaining a substantially constant zoom setting throughout an operation with the operation extending to model construction and deployment for a selected camera view of the single security surveillance camera; c) creating, by at least one computer processor, a nodelist comprising a collection of nodes, wherein each node comprises an abstraction of coordinates of a landmark in the at least one underlying terrain along with corresponding image pixel coordinates of the landmark; d) generating, by at least one computer processor, an element list by inputting the nodelist, to at least one triangular mesh generation operation, in a camera space, the at least one triangular mesh generation operation yielding triangular elements with non-intersecting sides; e) computing, by at least one computer processor, ray equations in three-dimensional space for each of the nodes associated with vertices of the triangular elements; and f) constructing, by at least one computer processor, the camera model for the selected camera view by scanning across each resolution cell of the associated camera image and determining for each pixel coordinates an encompassing triangular element in the element list, and for each such encompassing triangular element located, deriving an equation for a ray associated with the virtual camera point and the pixel coordinates of that resolution cell encompassed by the encompassing triangular element, through at least one interpolation of rays associated with vertices of the encompassing element, with the at least one interpolation being influenced by proximity of the resolution cell to the vertices of the encompassing triangular element.
-
Specification