Rotation state estimation apparatus, method thereof, and program

1 . A spin state estimation device, wherein an absolute value of w is an integer of 2 or more and u is a unit time, comprising processing circuitry configured to execute operations comprising:

obtaining a first object image, wherein the first object image represents an image of an object at a certain time point t obtained from an input video of a plurality of frames in time-series;

obtaining a second object image at a time point t+w·u from the input video;

creating, based on the first object image, a target estimation image, wherein the target estimation image represents another image of the object at the time point t+w·u by applying a simulated rotation by w unit time according to a hypothesis of spin state to the object in the first object image;

calculating, for each of a plurality of hypotheses of spin states and a plurality of w, a likelihood of the target estimation image for the second object image; and

estimating a spin state of the object by selecting a specific hypothesis of spin state and a specific w in which the calculated likelihood becomes high from among the plurality of hypotheses of spin states and the plurality of w, wherein the likelihood of the target estimation image becoming high represents a likelihood representing a predetermined condition.

2 . The spin state estimation device according to claim 1 , wherein

K is an integer of 2 or more, and the processing circuitry uses target estimation images which are images of an object of time points t 1 +w·u, t 2+w·u, . . . , t K+w·u obtained by applying the simulated rotation by w unit time according to the hypothesis of spin state to the object in object images at time points t 1 , t 2 , . . . , t k and object images at the time points t 1 +w·u, t 2 +w·u, . . . , t K +w·u obtained from the input video, to estimate the spin state of the object by respectively selecting the specific hypothesis of spin state and w in which the likelihood of the target estimation image becomes high from among the plurality of hypotheses of spin states and the plurality of w.

3 . The spin state estimation device according to claim 1 , wherein

the processing circuitry

estimates information corresponding to a provisional spin amount of the object, and

estimates the spin state of the object by respectively selecting the specific hypothesis of spin state and w in which the likelihood of the target estimation image becomes high from among the plurality of hypotheses of spin states and a search range of w based on information corresponding to the provisional spin amount.

4 . The spin state estimation device according to claim 1 , wherein

the hypothesis of spin state represents information corresponding to a spin shaft of the object and information corresponding to a spin amount of the object, and

the processing circuitry obtains information corresponding to at least one of the spin shaft of the object and the spin amount per unit time as the spin state of the object on the basis of information r w corresponding to the spin shaft of the object, information θ w corresponding to the spin amount of the object, and w represented by a selected hypothesis of spin state.

5 . The spin state estimation device according to claim 1 , wherein

the hypothesis of spin state represents information corresponding to a spin shaft of the object and information corresponding to a spin amount of the object, and

the processing circuitry

obtains information including information corresponding to the spin amount per unit time and not including information corresponding to the spin shaft of the object as the spin state of the object and/or outputs an effect that information corresponding to the spin shaft of the object cannot be estimated, when the spin amount per unit time obtained based on information θ w and w corresponding to the spin amount of the object represented by the selected hypothesis of spin state is π+2 nπ.

6 . The spin state estimation device according to claim 1 , wherein

the predetermined condition comprises the likelihood indicating a maximum value.

7 . The spin state estimation device according to claim 1 , wherein

the predetermined condition comprises the likelihood being equal to or more than a threshold value.

8 . The spin state estimation device according to claim 1 , wherein

the predetermined condition comprises an order of the likelihood being equal to or more than a reference order in descending order of the likelihood.

9 . A non-transitory computer-readable recording medium storing a program for causing a computer to function as a spin state estimation device according to claim 1 .

10 . A spin state estimation method, wherein an absolute value of w is an integer of 2 or more and u is a unit time, comprising:

obtaining a first object image, wherein the first object image represents an image of an object at a time point t obtained from an input video of a plurality of frames in time-series;

obtaining a second object image at a time point t+w·u from the input video;

creating, based on the first object image, a target estimation image, wherein the target estimation image represents another image of the object at the time point t+w·u by applying a simulated rotation by w unit time according to a hypothesis of spin state to the object in the first object image;

calculating, for each of a plurality of hypotheses of spin states and a plurality of w, a likelihood of the target estimation image for the second object image; and

estimating a spin state of the object by selecting a specific hypothesis of spin state and a specific w in which the calculated likelihood becomes high from among the plurality of hypotheses of spin states and the plurality of w, wherein the likelihood of the target estimation image becoming high represents a likelihood representing a predetermined condition.