US 12,080,023 B2
Trajectory linking apparatus, trajectory linking method, and non-transitory computer readable medium storing program
Yukie Ebiyama, Tokyo (JP); and Ryoma Oami, Tokyo (JP)
Assigned to NEC CORPORATION, Tokyo (JP)
Appl. No. 17/618,559
Filed by NEC Corporation, Tokyo (JP)
PCT Filed Jun. 25, 2019, PCT No. PCT/JP2019/025138
§ 371(c)(1), (2) Date Dec. 13, 2021,
PCT Pub. No. WO2020/261378, PCT Pub. Date Dec. 30, 2020.
Prior Publication US 2022/0245855 A1, Aug. 4, 2022
Int. Cl. G06T 7/73 (2017.01); G06V 10/62 (2022.01); G06V 10/74 (2022.01); G06V 10/77 (2022.01)
CPC G06T 7/73 (2017.01) [G06V 10/62 (2022.01); G06V 10/761 (2022.01); G06V 10/7715 (2022.01); G06T 2207/30241 (2013.01); G06V 2201/07 (2022.01)] 10 Claims
OG exemplary drawing
 
1. A trajectory linking apparatus comprising:
at least one memory storing instructions; and
at least one processor configured to execute the instructions to:
receive, for each of a plurality of time segments, trajectory information representing a plurality of trajectories indicating respective movement paths of a plurality of objects detected from image data of the each time segment, and indicating whether or not each of the trajectories has a feature value permitting object identification;
generate a linked trajectory having the feature value by linking a first trajectory included in first trajectory information obtained from the image data in a first time segment to a second trajectory included in second trajectory information obtained from the image data in a second time segment different from the first time segment, the first trajectory and the second trajectory each having the feature value permitting object identification;
generate an interpolated trajectory by linking a third trajectory included in third trajectory information obtained from the image data in a third time segment between the first and second time segments to the first and second trajectories, the third trajectory not having the feature value permitting object identification as a result of object occlusion; and
track a corresponding object from the image data in the first time segment, through the image data in the third time segment, to the image data in the second time segment, based on the interpolated trajectory, wherein the third trajectory does not have the feature value permitting object identification.