Method and system for estimating the three dimensional position of an object in a three dimensional physical space

A system for estimating the three dimensional position of an object, comprising:

• at least one image capturing device for capturing a plurality of images of a human form within a three dimensional (3D) physical space, each of said plurality of images captured from a different viewpoint location of said human form using said at least one image capturing device calibrated within said 3D physical space in a manner that does not require said human form to be close to a frontal position during said capturing of said plurality of images, each of said plurality of images is of said human form not in a frontal position;
• and
• a processor to:
• extract a plurality of silhouettes of said human form from said plurality of images, each of said plurality of silhouettes is from said at least one of said plurality of images of said human form not in said frontal position;
• obtain a plurality of head contours of a head of said human form from said plurality of silhouettes, wherein a contour of said plurality of contours corresponds to at least a portion of said human form and at least one of said contours is not an elliptical shape;
• determine a location of said head within said 3D physical space using an elliptical model of said head based on said plurality of head contours;
• project said elliptical model as an ellipse into a corresponding image taken of said human form from a corresponding viewpoint's location;
• minimize an error difference between said ellipse and a corresponding head contour from said corresponding viewpoint;
• and
• measure said error difference between said ellipse and said corresponding head contour.

A computer system comprising:

• a processor;
• a bus coupled to said processor;
• and
• a non-transitory computer-readable storage medium coupled to said bus containing instructions that when implemented cause said processor to implement a method for estimating the three dimensional position of an object, comprising:
• capturing a plurality of images of a human form within a three dimensional (3D) physical space in a manner that does not require said human form to be close to a frontal position during said capturing of said plurality of images, each of said plurality of images captured from a different viewpoint location of said human form using at least one image capturing device calibrated within said 3D physical space, each of said plurality of images is of said human form not in a frontal position;
• extracting a plurality of silhouettes of said human form from said plurality of images, wherein each of said plurality of silhouettes is from said at least one of said plurality of images of said human form not in said frontal position;
• obtaining a plurality of contours of an object of said human form from said plurality of silhouettes, wherein a contour of said plurality of contours corresponds to at least a portion of said human form and wherein at least one of said contours is not an elliptical shape;
• and
• determining a location of said object within said 3D physical space using an object model of said object based on said plurality of contours, wherein said determining a location comprises:
• projecting said object model as an ellipse into a corresponding image taken of said human form from a corresponding viewpoint's location;
• and
• minimizing an error difference between said ellipse and a corresponding contour of said object from said corresponding viewpoint, wherein said minimizing an error different comprises:
• measuring said error difference between said ellipse and said corresponding contour.

A non-transitory computer-readable storage medium containing program instructions that implement a method for estimating the three dimensional position of an object, comprising:

• capturing a plurality of images of a human form within a three dimensional (3D) physical space in a manner that does not require said human form to be close to a frontal position during said capturing of said plurality of images, each of said plurality of images captured from a different viewpoint location of said human form using at least one image capturing device calibrated within said 3D physical space, each of said plurality of images is of said human form not in a frontal position;
• extracting a plurality of silhouettes of said human form from said plurality of images, wherein each of said plurality of silhouettes is from at least one of said plurality of images of said human form not in said frontal position;
• obtaining a plurality of head contours of a head of said human form from said plurality of silhouettes, wherein a contour of said plurality of contours corresponds to at least a portion of said human form and wherein at least one of said contours is not an elliptical shape;
• determining a location of said object within said 3D physical space using an object model of said object based on said plurality of contours, wherein said determining a location comprises:
• projecting said object model as an ellipse into a corresponding image taken of said human form from a corresponding viewpoint's location;
• and
• minimizing an error difference between said ellipse and a corresponding contour of said object from said corresponding viewpoint, wherein said minimizing an error different comprises:
• measuring said error difference between said ellipse and said corresponding contour.