Individual identifying device
An imaging unit, an extraction unit, a feature amount pair generation unit, and an imaging parameter adjustment unit are included. The imaging unit acquires images obtained by imaging each of N (N≥3) types of objects a plurality of times by setting a value of a specific imaging parameter, among a plurality of types of imaging parameters, as a certain candidate value and changing a value of the remaining imaging parameter. The extraction unit extracts a feature amount from each of the images. The feature amount pair generation unit generates, as a first feature amount pair for each of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from images of objects of the same type, and generates, as a second feature amount pair for every combination of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a images of objects of the different types. The imaging parameter adjustment unit generates a first distribution that is a distribution of collation scores of the first feature amount pairs, generates a second distribution that is a distribution of collation scores of the second feature amount pairs, and on the basis of a degree of separation between the first distribution and the second distribution, determines the propriety of adopting the candidate value.
1. An individual identifying device comprising:
a memory containing program instructions; and
a processor coupled to the memory, wherein the processor is configured to execute the program instructions to:
a) set a value of a specific imaging parameter, among a plurality of types of imaging parameters, as a certain candidate value;
(b) acquire a plurality of images obtained by imaging each of N types of object a plurality of times by changing a value of a remaining imaging parameter, where N is a value greater than or equal to three;
(c) extract a feature amount from each of the plurality of the images;
(d) generate, as a first feature amount pair for each of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a plurality of images of objects of a same type, and generate as a second feature amount pair for every combination of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a plurality of images of objects of different types; and
(e) generate a first distribution that is a distribution of collation scores of a plurality of the first features amount pairs, generate a second distribution that is a distribution of collation scores of a plurality of the second feature amount pairs, and on a basis of a degree of separation between the first distribution and the second distribution, determine propriety of adopting the candidate value.
2. The individual identifying device according to claim 1 , wherein the processor is further configured to execute the instructions to
(f) after changing the value of the specific imaging parameter to another candidate value different from the candidate value, repeatedly perform the (b) to the (e).
3. The individual identifying device according to claim 2 , wherein
the (e) includes selecting, from among a plurality of the candidate values, one candidate value in which the degree of separation between the first distribution and the second distribution becomes equal to or larger than a preset threshold, as the value of the specific imaging parameter.
4. The individual identifying device according to claim 3 , wherein the processor is further configured to execute the instructions to
register the feature amount extracted from an image acquired by imaging a registration object under an imaging condition defined by the selected value of the specific imaging parameter in a storage device, in association with supplementary information of the registration object.
5. The individual identifying device according to claim 4 , wherein the processor is further configured to execute the instructions to
collate the feature amount extracted from an image acquired by imaging a collation object under the imaging condition defined by the selected value of the specific imaging parameter with the feature amount stored in the storage device.
6. The individual identifying device according to claim 2 , wherein
the (e) includes selecting, from among a plurality of the candidate values, a candidate value in which the degree of separation between the first distribution and the second distribution becomes highest, as the value of the specific imaging parameter.
7. The individual identifying device according to claim 1 , wherein
in the (b), a distance from a coaxial vertical illuminator to a surface of the object as the specific imaging parameter, the coaxial vertical illuminator irradiating the surface of the object with illumination light from a light source along an optical axis of a camera lens.
8. The individual identifying device according to claim 1 , wherein
in the (b), a distance from a ring illuminator that illuminates a surface of the object to the surface of the object is used as the specific imaging parameter.
9. The individual identifying device according to claim 1 , wherein
in the (b) a distance from a dome illuminator that illuminates a surface of the object to the surface of the object is used as the specific imaging parameter.
10. An individual identifying method comprising:
acquiring a plurality of images obtained by imaging each of N types of objects a plurality of times by setting a value of a specific imaging parameter, among a plurality of types of imaging parameters, as a certain candidate value and changing a value of a remaining imaging parameter, where N is a value greater than or equal to three;
extracting a feature amount from each of the plurality of the images;
generating, as a first feature amount pair for each of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a plurality of images of objects of a same type, and generating, as a second feature amount pair for every combination of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a plurality of images of objects of different types; and
generating a first distribution that is a distribution of collation scores of a plurality of the first feature amount pairs, generating a second distribution that is a distribution of collation scores of a plurality of the second feature amount pairs, and on a basis of a degree of separation between the first distribution and the second distribution, determining propriety of adopting the candidate value.
11. A non-transitory computer-readable medium storing a program comprising instructions for causing a computer execute processing to:
acquire a plurality of images obtained by imaging each of N types of objects a plurality of times by setting a value of a specific imaging parameter, among a plurality of types of imaging parameters, as a certain candidate value and changing a value of a remaining imaging parameter, where N is a value greater than or equal to three;
extract a feature amount from each of the plurality of the images;
generate, as a first feature amount pair for each of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a plurality of images of objects of a same type, and generate, as a second feature amount pair for every combination of the N types of objects, a feature amount pair in which two feature amounts constituting the feature amount pair are extracted from a plurality of images of objects of different types; and
generate a first distribution that is a distribution of collation scores of a plurality of the first feature amount pairs, generate a second distribution that is a distribution of collation scores of a plurality of the second feature amount pairs, and on a basis of a degree of separation between the first distribution and the second distribution, determine propriety of adopting the candidate value.