Detecting multiple moving objects in crowded environments with coherent motion regions
First Claim
1. An apparatus for identifying at least one discrete moving object within a series of video images, said apparatus comprising:
- an image recording device that stores a time series of video images; and
an image analysis device that is configured to execute a program of machine-executable instructions to identify at least one moving object in said time series of video images, wherein said program includes the steps of;
identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size;
calculating a trajectory similarity factor for each pair of feature point tracks said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks; and
generating a trajectory similarity matrix S, wherein said trajectory similarity matrix S is a Z×
Z matrix and Z is a total number of feature point tracks in said time series of video images, and wherein each element Sij in an i-th row and a j-th column in said trajectory similarity matrix S is given by S(i,j) that represents a trajectory similarity factor between an i-th feature point track and a j-th feature point track for all combinations of values for i and j.
3 Assignments
0 Petitions
Accused Products
Abstract
Coherent motion regions extend in time as well as space, enforcing consistency in detected objects over long time periods and making the algorithm robust to noisy or short point tracks. As a result of enforcing the constraint that selected coherent motion regions contain disjoint sets of tracks defined in a three-dimensional space including a time dimension. An algorithm operates directly on raw, unconditioned low-level feature point tracks, and minimizes a global measure of the coherent motion regions. At least one discrete moving object is identified in a time series of video images based on the trajectory similarity factors, which is a measure of a maximum distance between a pair of feature point tracks.
29 Citations
52 Claims
-
1. An apparatus for identifying at least one discrete moving object within a series of video images, said apparatus comprising:
-
an image recording device that stores a time series of video images; and an image analysis device that is configured to execute a program of machine-executable instructions to identify at least one moving object in said time series of video images, wherein said program includes the steps of; identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size; calculating a trajectory similarity factor for each pair of feature point tracks said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks; and generating a trajectory similarity matrix S, wherein said trajectory similarity matrix S is a Z×
Z matrix and Z is a total number of feature point tracks in said time series of video images, and wherein each element Sij in an i-th row and a j-th column in said trajectory similarity matrix S is given by S(i,j) that represents a trajectory similarity factor between an i-th feature point track and a j-th feature point track for all combinations of values for i and j. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)
-
-
16. An apparatus for identifying at least one discrete moving object within a series of video images, said apparatus comprising:
-
an image recording device that stores a time series of video images; and an image analysis device that is configured to execute a program of machine-executable instructions to identify at least one moving object in said time series of video images, wherein said program includes the steps of; identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size; calculating a trajectory similarity factor for each pair of feature point tracks said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks, wherein a trajectory similarity factor between an i-th feature point track and a j-th feature point track is given by the formula, S(i,j)=exp[−
α
×
DQ(i,j)−
β
×
DVQ(i,j))], wherein S(i,j) is said trajectory similarity factor between said i-th feature point track and said j-th feature point track, α and
β
are positive constants, DQ(i,j) is a quantity representing a first statistical quantity of a set of non-linear parameters derived from distances between said i-th feature point track and said j-th feature point track in said time series of video images, and DVQ(i,j) is a second statistical quantity of said set of non-linear parameters for all combination of values for i and j. - View Dependent Claims (17, 18, 19, 20)
-
-
21. A method for identifying at least one discrete moving object within a series of video images, said method comprising:
-
storing a time series of video images in an image recording device; identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size; calculating a trajectory similarity factor for each pair of feature point tracks in said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks; generating a trajectory similarity matrix S, wherein said trajectory similarity matrix S is a Z×
Z matrix and Z is a total number of feature point tracks in said time series of video images, and wherein each element Si,j in an i-th row and a j-th column in said trajectory similarity matrix S is given by S(i,j) that represents a trajectory similarity factor between an i-th feature point track and a j-th feature point track for all combinations of values for i and j; andidentifying at least one discrete moving object in said time series of video images based on said trajectory similarity factors. - View Dependent Claims (22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35)
-
-
36. A method for identifying at least one discrete moving object within a series of video images, said method comprising:
-
storing a time series of video images in an image recording device; identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size; calculating a trajectory similarity factor for each pair of feature point tracks in said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks; identifying at least one discrete moving object in said time series of video images based on said trajectory similarity factors, wherein a trajectory similarity factor between an i-th feature point track and a j-th feature point track is given by the formula, S(i,j)=exp[−
α
×
DQ(i,j)−
β
×
DVQ(i,j))], wherein S(i,j) is said trajectory similarity factor between said i-th feature point track and said j-th feature point track, α and
β
are positive constants, DQ(i,j) is a quantity representing a first statistical quantity of a set of non-linear parameters derived from distances between said i-th feature point track and said j-th feature point track in said time series of video images, and DVQ(i,j) is a second statistical quantity of said set of non-linear parameters for all combination of values for i and j. - View Dependent Claims (37, 38, 39, 40)
-
-
41. A machine-readable data storage device embodying a program of machine-executable instructions to identify at least one discrete moving object, wherein said program includes the steps of:
-
identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size; calculating a trajectory similarity factor for each pair of feature point tracks said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks; generating a trajectory similarity matrix S, wherein said trajectory similarity matrix S is a Z×
Z matrix and Z is a total number of feature point tracks in said time series of video images, and wherein each element Si,j in an i-th row and a j-th column in said trajectory similarity matrix S is given by S(i,j) that represents a trajectory similarity factor between an i-th feature point track and a j-th feature point track for all combinations of values for i and j; andidentifying at least one discrete moving object in said time series of video images based on said trajectory similarity factors. - View Dependent Claims (42, 43, 44, 45, 46, 47, 48, 49, 50)
-
-
51. A machine-readable data storage device embodying a program of machine-executable instructions to identify at least one discrete moving object, wherein said program includes the steps of:
-
identifying a plurality of feature point tracks in said time series of video images, wherein for each video image of said time series of video images, all feature points belonging to a same feature point track are located within a polygon of a predefined size; calculating a trajectory similarity factor for each pair of feature point tracks said plurality of feature point tracks, wherein each trajectory similarity factor is a measure of a maximum distance between a pair of feature point tracks; identifying at least one discrete moving object in said time series of video images based on said trajectory similarity factors, wherein a trajectory similarity factor between an i-th feature point track and a j-th feature point track is given by the formula, S(i,j)=exp [−
α
×
DQ(i,j) −
β
×
DVQ(i,j)) ], wherein S(i,j) is said trajectory similarity factor between said i-th feature point track and said j-th feature point track, α and
β
are positive constants, DQ(i,j) is a quantity representing a first statistical quantity of a set of non-linear parameters derived from distances between said i-th feature point track and said j-th feature point track in said time series of video images, and DVQ(i,j) is a second statistical quantity of said set of non-linear parameters for all combination of values for i and j. - View Dependent Claims (52)
-
Specification