Feature grouping in moving bodies in an image

Claim 2

An imaging system for determining the number of independent moving bodies in image frames containing a plurality of moving bodies, associating points in the image with respective bodies and determining the trajectories of the moving bodies comprising: means for obtaining a sequence of images containing a plurality of moving bodies; feature detector means for obtaining image data related to identified points arising from the moving bodies in said sequence of images; computing means for receiving said image data related to identified points, said computing means: forming a matrix W where column vector w i of matrix W includes the coordinates of each ith point; representing the set of all points of M=diag(w 1 ,w 2 , . . . ,w s ) where w p , p=1, . . . , s are point coordinates and there are s bodies and each body has a set of n p feature points; forming a block transformation matrix T having entries T jp ,j=1, . . . , m and p=1, . . . , s where T jp is the pth object in the jth frame subject to a transformation/projection; representing the observed data, matrix D, as D=TM where the upper bound on the rank of D is 4 s and the lower bound is s where s is the number of independent moving bodies; forming the echelon form of matrix D; and providing output data commensurate with the number of independent bodies in the image and with which points are associated with respective bodies from the echelon form of matrix D where columns having non-zero elements in the same row correspond to points belonging to same rigid body if matrix T has full rank and where the rank of echelon form of matrix D is related to a lower bound on the number of independent bodies in the image, and processor means receiving said output data for determining the trajectories of each moving body in said sequence of images.

Claim 3

An imaging system as set forth in claim 1, where when matrix T is not of full rank, said forming a block transformation matrix T comprises deleting a column of matrix T, subtracting a multiple of a row of W from other rows and deleting that row from W.

Claim 4

An imaging system as set forth in claim 1, wherein forming the echelon form of matrix D includes selecting a tolerance value.

Claim 5

An imaging system as set forth in claim 1, wherein forming the echelon form of matrix D includes selecting a rank tolerance value and a grouping tolerance value.

Claim 6

An imaging system as set forth in claim 1, wherein forming the echelon form of matrix D includes applying Gauss-Jordan elimination to rows of matrix D.

Claim 7

An imaging system as set forth in claim 1, wherein forming the echelon form of matrix D includes includes column scaling and shifting of matrix D.

Claim 8

An imaging system as set forth in claim 1, wherein forming the echelon form of matrix D includes row scaling and shifting of matrix D.

Claim 9

An imaging system as set forth in claim 2, wherein forming the echelon form of matrix D includes selecting a tolerance value.

Claim 10

An imaging system as set forth in claim 2, wherein forming the echelon form of matrix D includes selecting a rank tolerance value and a grouping tolerance value.

Claim 11

An imaging system as set forth in claim 2, wherein forming the echelon form of matrix D includes applying Gauss-Jordan elimination to rows of matrix D.

Claim 12

An imaging system as set forth in claim 2, wherein forming the echelon form of matrix D includes includes column scaling and shifting of matrix D.

Claim 13

An imaging system as set forth in claim 2, wherein forming the echelon form of matrix D includes row scaling and shifting of matrix D.

Claim 14

An imaging system as set forth in claim 1, further comprising means for supporting said imaging system and said processor means determining whether any of the moving bodies will collide with said means for supporting said imaging system.

Claim 15

An imaging system set forth in claim 1, where said processor means identifies moving bodies traveling relative to said means for obtaining a sequence of images.

Claim 16

An imaging system as set forth in claim 1, where said processor means identifies the motion of the moving bodies relative to said means for obtaining a sequence of images.

Claim 17

An imaging system including a camera, a feature detector, a computer, and a processor for determining the number of independent moving bodies in image flames containing a plurality of moving bodies and associating points in the image with respective bodies for determining the trajectories of the moving bodies comprising; a camera for recording a sequence of image frames of a scene containing a plurality of moving bodies; a feature detector for obtaining image data related to identified points arising from the moving bodies in said sequence of image frames recorded by the camera; a computer for receiving said image data related to identified points, said computer: forming a matrix W where column vector w i of matrix W includes the coordinates of each ith point; representing the set of all points by M=diag(w i , w 2 , . . . , w s ) where w p ,p=1, . . . , s are point coordinates and there are s bodies and each body has a set of n p feature points; forming a block transformation matrix T having entries T jp ,j=1, . . . , m and p=1, . . . , s, where T jp is the pth object in the jth frame subject to a transformation/projection; representing the observed data, matrix D, as D=TM where the upper bound on the rank of D is 4 s and the lower bound is s, where s is the number of independent moving bodies; forming the echelon form of matrix D; and providing output data commensurate with the number of independent bodies in the image and with which points are associated with respective bodies from the echelon form of matrix D where columns having non-zero elements in the same row correspond to points belonging to same rigid body if matrix T has full rank and where the rank of echelon form of matrix D is related to a lower bound on the number of independent bodies in the image, and a processor for receiving said output data and determining the trajectories of each moving body in the sequence of image frames.

Claim 18

An imaging system including a camera, a feature detector, a computer, and a processor for determining the number of independent moving bodies in image flames containing a plurality of moving bodies and associating points in the image with respective bodies for determining the trajectories of the moving bodies comprising; a camera for recording a sequence of image frames of a scene containing a plurality of moving bodies; a feature detector for obtaining image data related to identified points arising from the moving bodies in said sequence of image frames recorded by the camera; a computer for receiving said image data related to identified points, said computer: forming a matrix W where column vector w i of matrix W includes the coordinates of each ith point; representing the set of all points by M=diag(w i , w 2 , . . . , w s ) where w p ,p=1, . . . , s are point coordinates and there are s bodies and each body has a set of n p feature points; forming a block transformation matrix T having entries T jp ,j=1, . . . , m and p=1, . . . , s, where T jp is the pth object in the jth frame subject to a transformation/projection; representing the observed data, matrix D, as D=TM where the upper bound on the rank of D is 4 s and the lower bound is s, where s is the number of independent moving bodies; forming the echelon form of matrix D; and providing output data commensurate with the number of independent bodies in the image and with which points are associated with respective bodies from the echelon form of matrix D where columns having non-zero elements in the same row correspond to points belonging to same rigid body if matrix T has full rank and where the rank of echelon form of matrix D is related to a lower bound on the number of independent bodies in the image, and a processor for receiving said output data and determining the trajectories of each moving body in the sequence of image frames.

Claim 19

An imaging system as set forth in claim 16, where when matrix T is not of full rank, said forming a block transformation matrix T comprises deleting a column of matrix T, subtracting a multiple of a row of W from other rows and deleting that row from W.

Claim 20

An imaging system as set forth in claim 16, wherein forming the echelon form of matrix D includes selecting a tolerance value.

Claim 21

An imaging system, a method as set forth in claim 16, wherein forming the echelon form of matrix D includes selecting a rank tolerance value and a grouping tolerance value.

Claim 22

An imaging system, a method as set forth in claim 16, wherein the step of forming the echelon form of matrix D includes applying Gauss-Jordan elimination to rows of matrix D.

Claim 23

An imaging system as set forth in claim 17, wherein forming the echelon form of matrix D includes selecting a tolerance value.

Claim 24

An imaging system as set forth in claim 17, wherein forming the echelon form of matrix D includes selecting a rank tolerance value and a grouping tolerance value.

Claim 25

An imaging system as set forth in claim 17, wherein forming the echelon form of matrix D includes applying Gauss-Jordan elimination to rows of matrix D.

Claim 26

An imaging system as set forth in claim 16, further comprising means for supporting said imaging system and said processor determining whether any of the moving bodies will collide with said means for supporting said imaging system.

Claim 27

An imaging system as set forth in claim 16, where said processor identifies moving bodies traveling relative to said camera.

Claim 28

An imaging system as set forth in claim 16, where said processor identifies the motion of the moving bodies relative to said camera.