×

Systems and methods for extracting information about objects from scene information

  • US 9,904,867 B2
  • Filed: 01/29/2017
  • Issued: 02/27/2018
  • Est. Priority Date: 01/29/2016
  • Status: Active Grant
First Claim
Patent Images

1. A method of generating information about one or more objects of interest in a scene, comprising:

  • a. providing a plurality of overlapping 2D (two-dimensional) images of a scene, wherein;

    i. the scene comprises one or more objects; and

    ii. the plurality of overlapping 2D images is derived from one or more single passive image capture devices;

    b. selecting an object from the scene, thereby providing a selected object of interest, wherein the selection is by a computer or a user;

    c. generating 2D image information from the plurality of overlapping 2D images of the scene, wherein at least some of the plurality of overlapping 2D images;

    i. include all or part of the selected object, andii. are overlapping as to all or part of the selected object, 

    thereby providing 2D image information incorporating all or part of the selected object;

    d. providing, by a computer, 3D (three-dimensional) information generated from the scene, wherein the 3D information from the scene incorporates information about all or part of the selected object, thereby providing 3D information that incorporates all or part of the selected object;

    e. generating projective geometry information by combining at least some of the 2D image information incorporating all or part of the selected object and at least some of the 3D information incorporating all or part of the selected object, thereby establishing a plurality of relationships between 3D data points derived from either or both of the 2D image information incorporating all or part of the selected object and the 3D image information incorporating all or part of the selected object, wherein one or more of;

    1. measurement information,2. dimensions, and3. geometric information, 

    is derivable for all or part of the selected object, and wherein the plurality of relationships are established by;

    i. performing an image segmentation or abstraction step on the 2D image information by partitioning and grouping image elements, thereby generating a plurality of segmented 2D image information, wherein at least some of the plurality of segmented 2D image information includes information about at least part of the selected object;

    ii. performing a clustering step on the 3D information incorporating all or part of the selected object to partition and group 3D data points present in the 3D information that belong together, wherein the 3D data points are associated with all or part of the selected object, thereby generating a plurality of clustered 3D information, wherein at least some of the plurality of clustered 3D information includes information about at least part of the selected object; and

    iii. performing a plurality of cross-validation steps on each of the segmented 2D image information and the clustered 3D information by processing each of the plurality of segmented 2D image information and the plurality of clustered 3D information along with the combined 2D image information and the 3D information.

View all claims
  • 2 Assignments
Timeline View
Assignment View
    ×
    ×