Rotation state estimation device, method and program
A technique that enables, even if a video image corresponding to one period is not obtained, estimating a rotational state of an object is provided. The rotational state estimation device includes: an object image generation unit 1 that generates object images, each of the object images being an image of an object, from an input video image; and a rotational state estimation unit 3 that estimates a rotational state of the object by selecting, from among a plurality of hypotheses of the rotational state, a hypothesis of the rotational state, a likelihood of an image of the object resulting from the object in the object image at a certain time being rotated for t c unit time based on the hypothesis of the rotational state being high, using the object image at a time t and the object image at a time t+t c where t c is a predetermined integer of no less than 1.
1. A rotational state estimation device comprising:
processing circuitry configured to
generate a first object image of an object and a second object image of the object, from an input video image of the object in rotation; and
estimate a number of rotations of the object in a predetermined time by selecting, among a plurality of hypotheses of the rotational state, a hypothesis of the rotational state using the first object image of the object at time t and the second object image of the object at a time t+t c , wherein t c is a predetermined integer of no less than one, wherein the selecting is performed based on a determination that a calculated hypothesis likelihood for the selected hypothesis has a maximum value of calculated hypothesis likelihoods of the plurality of hypotheses, and wherein the hypothesis likelihood for the selected hypothesis indicates a highest likelihood of a rotational state of the second object image of the object matching a rotational state of a third object image of the object at a time t+t c , and wherein the third object image of the object represents the object in the first object image at the time t being rotated for t c unit time based on the hypothesis of the rotational state.
2. The rotational state estimation device according to claim 1 , wherein the processing circuitry configured to estimate the rotational state of the object by selecting, from among a plurality of hypotheses of the rotational state, a hypothesis of the rotational state, a likelihood of the third object image of the object resulting from the object in respective object images of the object at times t 1 , t 2 , . . . , t K being rotated for t c unit time based on the hypothesis of the rotational state being high, using the respective object images at the times t 1 , t 2 , . . . , t K and respectively corresponding object images at times t 1 +t c , t 2 +t c , . . . , t k +t c .
3. The rotational state estimation device according to claim 1 , wherein the processing circuitry configured to repeatedly perform processing for, for each of the plurality of hypotheses of the rotational state, calculating a likelihood of an image of the object resulting from the object in respective object images at the times t 1 , t 2 , . . . t K being rotated for t c unit time based on the hypothesis of the rotational state, and processing for newly generating a plurality of likely hypotheses of the rotational state based on the calculated likelihoods.
4. The rotational state estimation device according to claim 3 , wherein the processing for newly generating a plurality of likely hypotheses of the rotational state based on the calculated likelihoods, the processing being performed by the processing circuitry, is processing for newly generating a plurality of hypotheses by repeating, a plurality of times, processing for determining a hypothesis from among the plurality of hypotheses of the rotational state in such a manner as to determine the hypothesis, the calculated likelihood of the hypothesis being larger, with a higher probability, and determining the rotational state having a value resulting from addition of a dynamically generated number to a value of the rotational state of the determined hypothesis as a new hypothesis.
5. A rotational state estimation method comprising:
an object image generation step of causing an object image generation unit to generate a first object image of an object and a second object image of the object, from an input video image of the object in rotation; and
a rotational state estimation step of causing a rotational state estimation unit to estimate a number of rotations of the object in a predetermined time by selecting, among a plurality of hypotheses of the rotational state, a hypothesis of the rotational state using the first object image of the object at time t and the second object image of the object at a time t+t c , wherein t c is a predetermined integer of no less than one, wherein the selecting is performed based on a determination that a calculated hypothesis likelihood for the selected hypothesis has a maximum value of calculated hypothesis likelihoods of the plurality of hypotheses, and wherein the hypothesis likelihood for the selected hypothesis indicates a highest likelihood of a rotational state of the second object image of the object matching a rotational state of a third object image of the object at a time t+t c , and wherein the third object image of the object represents the object in the first object image at the time t being rotated for t c unit time based on the hypothesis of the rotational state.
6. A non-transitory computer readable medium that stores a program for making a computer perform the respective steps of the rotational state estimation method according to claim 5 .