Analysis apparatus and analysis method

An analysis apparatus, comprising:

• imager circuitry configured to record an image of an object person performing a first operation from a top-down view above the object person and acquire depth information from the top-down view above the object person;
• and
• a processor configured to:
• extract a head shape of the object person from the image using the depth information,
• identify the object person based on the head shape;
• a storage circuitry configured to store data that includes head shapes of a plurality of persons from a first person to an Nth person, N is an integer equal to or greater than 2, the storage circuitry further configured to store a plurality of identifiers from a first identifier to an Mth identifier, the Mth identifier being used to discriminate the Mth person from the other persons, M is an integer not less than 1 and not more than N,
• wherein the processor is further configured to:
• calculate each distance between the extracted head shape of the object person and each of the head shapes of the persons from the first person to the Nth person,
• calculate respective margins between the calculated distances and the identifiers from the first identifier to the Mth identifier, and
• determine the person corresponding to the identifier which the calculated margin is highest as the object person.

An analysis method comprising:

• a first step of acquiring first information based on motion of an object person, the first information with a first time length between a first time and a second time, the object person performing a first operation repeatedly between the first time and the second time;
• a second step of extracting a plurality of part information with a second time length from the first information at mutually-different times, the second time length being shorter than the first time length;
• a third step of extracting a plurality of first comparison information with the second time length from the first information, the plurality of third information being extracted between the first time and the second time at first time intervals;
• a fourth step of calculating distances between one of the plurality of part information and each of the plurality of first comparison information create first correlation information, the first correlation information including distances between the one of the plurality of part information and the first information at times respectively corresponding to the plurality of first comparison information;
• a fifth step of temporally setting a plurality of candidate points after a third time length from the first time;
• a sixth step of creating a plurality of third comparison information, the plurality of the third comparison information having peak at the respective candidate points;
• a seventh step of calculating a second correlation coefficient between each of the plurality of the third comparison information and the first correlation information;
• an eighth step of setting the candidate point which the second correlation coefficient is highest as a tentative similarity point;
• a ninth step of temporally repeating the set of other candidate points after the third time length from the last tentative similarity point and executing the sixth step to the eighth step for the other candidate points to set other tentative similarity points;
• a tenth step of calculating a first correlation coefficient between the first correlation information and second comparison information that is created based on the plurality of tentative similarity points,
• repeating the fourth step to the tenth step for each of the other mutually-different part information extracted in the second step, determining one group of the plurality of tentative similarity points corresponding to the first correlation information which the first correlation coefficient is highest, and calculating time intervals between the one group of the plurality of tentative similarity points;
• an eleventh step of acquiring depth information by recording the object person;
• a twelfth step of extracting a head shape of the object person from the depth information;
• a thirteenth step of using data that relates to head shapes of a plurality of persons from a first person to an Nth person and a plurality of identifiers from a first identifier to an Mth identifier so as to calculate distances between the extracted head shape of the object person and the head shapes of the plurality of persons, N is an integer equal to or greater than 2, the Mth identifier being used to discriminate the Mth person from other persons, M is an integer not less than 1 and not more than N;
• a fourteenth step of calculating respective margins between the calculated distances and the plurality of identifiers from the first identifier to the Nth identifier;
• and
• a fifteenth step of determining the person corresponding to the identifier which the calculated margin is highest as the object person.