Device and method for detecting finger position
A finger position detection device includes a processor; and a memory which stores a plurality of instructions, which when executed by the processor, cause the processor to execute, detecting a hand region that includes a plurality of fingers from an image captured by an image capturing unit that captures an image of the plurality of fingers from in front of the fingers; dividing a first region that is at least a part of the hand region into a plurality of first cells of a first size and extracting, from each of the plurality of first cells, a first feature value that represents a feature of a position of one of the plurality of fingers in a depth direction with respect to the image capturing unit; dividing a second region that is at least a part of the hand region into a plurality of second cells of a second size.
1. A finger position detection device comprising:
a processor; and
a memory which stores a plurality of instructions, which when executed by the processor, cause the processor to execute,
detecting a hand region that includes a plurality of fingers from an image captured by an image capturing unit that captures an image of the plurality of fingers from in front of the fingers;
dividing a first region that is at least a part of the hand region into a plurality of first cells of a first size and extracting, from each of the plurality of first cells, a first feature value that represents a feature of a position of one of the plurality of fingers in a depth direction with respect to the image capturing unit;
dividing a second region that is at least a part of the hand region into a plurality of second cells of a second size smaller than the first size and extracting, from each of the plurality of second cells, a second feature value that represents a feature of a position of one of the plurality of fingers in a finger arrangement axis along which the plurality of fingers are arranged;
detecting a finger position in the depth direction by inputting the first feature value into a first classifier that represents a relationship between the first feature value and a finger position in the depth direction; and
detecting a finger position in the finger arrangement axis by inputting the second feature value into a second classifier that represents a relationship between the second feature value and a finger position in the finger arrangement axis.
2. The finger position detection device according to claim 1 , further comprising:
a first-direction feature value extraction unit that sets a region above a given position in a height direction in the hand region as the first region, and
a second-direction feature value extraction unit that sets a region at or below the given position in the hand region as the second region.
3. The finger position detection device according to claim 1 , wherein the first feature value is a cumulative value of luminance gradients in each of a plurality of directions in each of the plurality of first cells.
4. The finger position detection device according to claim 1 , wherein the second feature value is a cumulative value of luminance gradients in each of a plurality of directions in each of the plurality of second cells.
5. The finger position detection device according to claim 1 , wherein the first classifier and the second classifier are boost classifiers.
6. A finger position detection method comprising:
detecting, by a computer processor, a hand region that includes a plurality of fingers from an image captured by an image capturing unit that captures an image of the plurality of fingers from in front of the fingers;
dividing a first region that is at least a part of the hand region into a plurality of first cells of a first size and extracting, from each of the plurality of first cells, a first feature value that represents a feature of a position of one of the plurality of fingers in a depth direction with respect to the image capturing unit;
dividing a second region that is at least a part of the hand region into a plurality of second cells of a second size smaller than the first size and extracting, from each of the plurality of second cells, a second feature value that represents a feature of a position of one of the plurality of fingers in a finger arrangement axis along which the plurality of fingers are arranged;
detecting a finger position in the depth direction by inputting the first feature value into a first classifier that represents a relationship between the first feature value and a finger position in the depth direction; and
detecting a finger position in the finger arrangement axis by inputting the second feature value into a second classifier that represents a relationship between the second feature value and a finger position in the finger arrangement axis.
7. The finger position detection method according to claim 6 ,
wherein, in the extracting the first feature value, a region above a given position in a height direction in the hand region is set as the first region, and
in the extracting the second feature value, a region at or below the given position in the hand region is set as the second region.
8. The finger position detection method according to claim 6 , wherein the first feature value is a cumulative value of luminance gradients in each of a plurality of directions in each of the plurality of first cells.
9. The finger position detection method according to claim 6 , wherein the second feature value is a cumulative value of luminance gradients in each of a plurality of directions in each of the plurality of second cells.
10. The finger position detection method according to claim 6 , wherein the first classifier and the second classifier are boost classifiers.