Method, device, and medium for determining three-dimensional position of skeleton using data acquired by multiple sensors
A recognition method executed by a processor, includes: acquiring positional relation between a plurality of sensors each of which senses a distance to an object; provisionally classifying an orientation of the object relative to each individual sensor included in the sensors into one of a plurality of classifications based on sensing data acquired by the individual sensor; calculating likelihood of each combination corresponding to the positional relation between the sensors based on a result of provisional classification of the orientation of the object relative to the individual sensor; and classifying the orientation of the object corresponding to each individual sensor in accordance with the calculated likelihood of each combination.
1. A recognition method executed by a processor, the recognition method comprising:
acquiring positional relation between a plurality of sensors each of which senses a distance to an object;
estimating a three-dimensional position of a skeleton of the object when viewed from each individual sensor included in the sensors based on sensing data acquired by the individual sensor and a skeleton recognition dictionary selected in accordance with an orientation of the object relative to the individual sensor, wherein
the estimating includes generating a site label image in which a site of the object is labeled based on a distance thereto included in the sensing data, in the site label image a boundary of the site of the object corresponding to a joint site, the site being expressed by a site label by using the skeleton recognition dictionary, and estimating the three-dimensional position of the skeleton of the object based on the site label in the generated site label image; and
determining three-dimensional position of the skeleton of the object based on a result of estimation of the three-dimensional position of the skeleton of the object when viewed from each individual sensor and the positional relation between the sensors.
2. The recognition method according to claim 1 , wherein the estimating includes estimating the three-dimensional position of the skeleton of the object by using data corresponding to a boundary of the site label in the sensing data.
3. The recognition method according to claim 1 , wherein the determining includes determining a barycenter of a boundary of the site label to be a joint position.
4. The recognition method according to claim 3 , wherein the determining includes calculating the barycenter of the boundary by using the result of estimation of the three-dimensional position of the skeleton when viewed from a sensor that has the shortest distance to the object among the sensors.
5. The recognition method according to claim 2 , wherein the estimating includes calculating dispersion of the data corresponding to the boundary for each individual sensor and estimating the three-dimensional position of the skeleton of the object by using sensing data obtained by randomly thinning or increasing the sensing data of each individual sensor by using reciprocal of the calculated dispersion.
6. The recognition method according to claim 3 , wherein the determining includes determining a position of the site based on determined joint positions.
7. The recognition method according to claim 3 , wherein the determining includes determining the barycenter of the site label to be a position of the site corresponding to the site label.