JP2017051241A - Specification device, specification method, and program for specifying pupil region of person in image - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a specification device for specifying a pupil region accurately.SOLUTION: A specification device includes: detection means for detecting a region including an eye of a person in an image indicated by image data; determination means for determining a value of a designated component from information indicating a color of a pixel for each pixel in the region; comparison means for comparing the value determined by the determination means with a threshold for each pixel in the region; and specification means for specifying a pupil region of the eye of the person from the result of the comparison by the comparison means.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a pupil specifying technique for specifying a pupil region of a person in an image.

  For example, as one method for estimating a person's emotion with respect to a visual stimulus, there is a method using a change in the pupil diameter of the person. Patent Document 1 discloses an apparatus for measuring the pupil diameter with high accuracy. However, the configuration described in Patent Document 1 has a strong constraint on the person to be measured, and the measurement apparatus is complicated and expensive. On the other hand, Non-Patent Document 1 discloses a configuration for specifying a human pupil diameter in an image from image data acquired by a Web camera.

JP 2014-79374 A

"Pupil detection from an eye image", [online], [searched on August 25, 2015], Internet <URL: http://opencv-code.com/tutorials/pupil-detection-from-an-eye- image />

  The method described in Non-Patent Document 1 detects, from image data, a rectangular area including human eyes in an image indicated by the image data, and whether or not the luminance value of a pixel in the rectangular area is equal to or greater than a threshold value. Thus, the pixels in the rectangular area are binarized, and thereby the pupil area is specified. The rectangular area generally includes the pupil, the iris, the sclera (white eye portion), and the skin. However, since the pupil is darker than the other areas, the pupil area can be specified by appropriately setting a threshold value. it can. However, it is desired to specify the pupil region with higher accuracy.

  The present invention provides a specifying device, a specifying method, and a program for specifying a pupil region with high accuracy.

  According to one aspect of the present invention, the specifying device is designated from detection means for detecting a region including human eyes in an image indicated by image data, and information indicating the color of the pixel for each pixel in the region. For each pixel in the region, a determination unit that determines a component value, a comparison unit that compares a value determined by the determination unit with a threshold value, and a pupil region of the human eye is identified from a comparison result by the comparison unit And a specifying means.

  According to the present invention, the pupil region can be specified with high accuracy.

The block diagram of the specific apparatus by one Embodiment. The figure which shows the luminance component and red component of a pupil area | region and an iris area | region. The block diagram of the specific apparatus by one Embodiment. The block diagram of the specific apparatus by one Embodiment.

  Hereinafter, exemplary embodiments of the present invention will be described with reference to the drawings. In addition, the following embodiment is an illustration and does not limit this invention to the content of embodiment. In the following drawings, components that are not necessary for the description of the embodiments are omitted from the drawings.

<First embodiment>
FIG. 1 is a configuration diagram of a specific apparatus according to the present embodiment. The eye area detection unit 1 determines a rectangular area including a human eye in each frame of an image indicated by input image data. In the present embodiment, the image data is moving image data, but may be still image data. In addition, a known face recognition technique can be used to determine a rectangular area including the eyes. The resolution processing unit 2 applies a super-resolution technique to increase the resolution of the rectangular area. If the resolution of the image data itself is high, the resolution processing unit 2 can be omitted. The color component determination unit 3 receives specification information indicating a color component designated by the user, and the color component determination unit 3 determines the color component specified by the specification information from the information indicating the color of each pixel in the rectangular area. Determine the value of and output. Threshold information indicating a threshold specified by the user is input to the binarization unit 4, and the binarization unit 4 is designated by designation information output from the color component determination unit 3 for each pixel in the rectangular area. It is determined whether the value of the selected color component is larger than the threshold value specified by the threshold information, and binarization information of each pixel in the rectangular area is output. Specifically, when the value of the color component of the pixel is larger than the threshold value, the binarization unit 4 outputs “1” for the pixel, and binarization is performed when the value of the color component of the pixel is equal to or less than the threshold value. The unit 4 outputs “0” for the pixel. The specifying unit 5 specifies the pupil region based on the pixel values in the rectangular region output from the binarizing unit 4.

  The color of the iris varies from person to person, such as dark brown, amber, green, and blue. In the case of a high-luminance iris such as blue, the pupil region can be specified by threshold determination based on the luminance.However, in the case of a low-luminance iris such as dark brown, the pupil region is accurately identified by the threshold determination based on luminance. It can not be identified. Furthermore, it is not preferable to measure the subject's eyes by illuminating the subject's eyes in order to make the subject less aware of the measurement for emotion estimation. Even in the case of an iris with high brightness, it is difficult to determine the pupil region. 2A and 2B show the average value of the luminance component value and the red component value of each pixel corresponding to the pupil and iris of a person with a dark brown iris. In addition, the measurement of FIG. 2 (A) and (B) is a value about the image image | photographed by illuminating the eyes of two test subjects. 2A and 2B, for example, in the case of a dark brown iris, the red component value has a larger difference between the pupil and iris values than the luminance component value. It can be seen that the use of the red component value is more advantageous for specifying the pupil region.

  As described above, in this embodiment, the color component used for specifying the pupil region is specified by the specification information. The designated color component can be, for example, any one of red, blue, and green. For example, as described above, in the case of a person having a dark brown iris, the pupil region can be specified with high accuracy by designating the red (R) component with the designation information. If the person has a blue iris, the pupil region can be accurately identified by designating the blue (B) component with the designation information. If the person has a green (G) iris, the pupil region can be accurately identified by designating the green component with the designation information. Further, instead of designating by so-called RGB color components, for example, a configuration may be adopted in which color difference components such as U or V of YUV are designated by designation information. Further, the specifying unit 5 can determine the size of the pupil from the specified pupil region. By performing this process on each frame of the moving image, it is possible to determine a change in pupil diameter, and thereby it is possible to estimate a human emotion with respect to a visual stimulus.

<Second embodiment>
Next, the second embodiment will be described focusing on the differences from the first embodiment. FIG. 3 is a configuration diagram of the specific apparatus according to the present embodiment. In the present embodiment, two components are designated as the designation information. For example, two components R (red) and G (green) can be designated. Also, for example, two components R (red) and Y (luminance component) can be designated. The color component determination unit 3 outputs the value of each component designated by the designation information for each pixel in the rectangular area including the eyes to the corresponding binarization unit. For example, if two components R and G are designated by the designation information, the color component determination unit 3 outputs the R component to the binarization unit 41 and outputs the G component to the binarization unit 42. Further, the user designates the threshold value for each component designated by the designation information using the threshold information. That is, the binarization unit 41 receives the R component threshold value as threshold information, and the binarization unit 42 receives the G component threshold value as threshold information. The binarization unit 41 and the binarization unit 42 output binarization information of each pixel in the rectangular area to the specifying unit 5 based on the threshold, and the specifying unit 5 includes the binarization unit 41 and the binarization unit. The pupil region is specified by the value of the pixel in the rectangular region output by 42.

  For example, the specifying unit 5 includes an overlapping portion of the pupil region specified by the binarization information output from the binarization unit 41 and the pupil region specified by the binarization information output from the binarization unit 42, that is, The region specified by the logical product (AND) of the binarization information output from the binarization units 41 and 42 can be determined as the pupil region. Alternatively, the specifying unit 5 is a portion in which the pupil region specified by the binarization information output by the binarization unit 41 and the pupil region specified by the binarization information output by the binarization unit 42 are combined, That is, the region specified by the logical sum (OR) of the binarization information output from the binarization units 41 and 42 can be determined as the pupil region. For example, when a shadow due to wrinkles occurs around the pupil, the accuracy of specifying the pupil region deteriorates. In this case, for example, the accuracy of specifying the pupil region can be improved by specifying the pupil region by logical product. On the other hand, when the vicinity of the pupil boundary becomes bright due to corneal reflection or the like, the binarized information of one component can determine that the portion that is a pupil is not a pupil. In this case, the accuracy of specifying the pupil region can be improved by specifying the pupil region by logical sum. In addition, the specifying unit 5 determines the average area of the pupil region specified by the binarization information output from the binarization unit 41 and the pupil region specified by the binarization information output from the binarization unit 42 as the pupil. It can also be specified as an area.

  As described above, in this embodiment, the pupil region is determined by designating two components. Thereby, the identification accuracy of the pupil region can be improved. At this time, as described in the first embodiment, at least one component is one of RGB or a color difference component. It is also possible to specify two or more components.

<Third embodiment>
Next, the third embodiment will be described focusing on the differences from the first embodiment. FIG. 4 is a configuration diagram of the specific apparatus according to the present embodiment. In this embodiment, the user does not specify the threshold value, but the binarization unit 4 adaptively sets the threshold value. For example, the binarization unit 4 obtains an average value of the component values designated by the designation information of all the pixels in the input rectangular area and sets this as M. Then, the threshold value Tb is obtained by Tb = α × M + β. Further, the binarization unit 4 first obtains a pixel corresponding to the iris or pupil by comparing the value of the component designated by the designation information of the pixel in the input rectangular area with a predetermined threshold value. The periphery of the iris is the so-called sclera (white eye portion), and the boundary between the iris and the sclera can be determined relatively easily. Therefore, the pixels corresponding to the iris or the pupil that is the area inside the iris can be determined relatively easily. Then, the binarization processing unit obtains the average value of the component values designated by the designation information of all the pixels corresponding to the iris and pupil, and sets this as M, and can also obtain the threshold value Tb by Tb = α × M + β. . Α and β are predetermined constants.

  As described above, in the present embodiment, the threshold value itself in the binarization process is obtained from the image data. Thereby, a threshold value more appropriate than the threshold value specified by the user can be set, and thus the accuracy of specifying the pupil region can be improved. In this embodiment as well, as in the second embodiment, the pupil region can be specified by the values of a plurality of components.

  The specific device according to the present invention can be realized by a program that causes a computer to operate as the specific device. These computer programs can be stored in a computer-readable storage medium or distributed via a network.

  3: color component determination unit, 4: binarization unit, 5: determination unit

Claims (11)

Detecting means for detecting a region including human eyes in the image indicated by the image data;
For each pixel in the region, determination means for determining a value of a designated component from information indicating the color of the pixel;
For each pixel in the region, a comparison unit that compares the value determined by the determination unit with a threshold value;
A specifying means for specifying a pupil region of the human eye from a comparison result by the comparing means;
An apparatus for identifying a pupil region, comprising:   The identification device according to claim 1, wherein the designated component is a red, green, or blue component.   The identification device according to claim 1, wherein the designated component is a color difference component. The designated component comprises two or more components;
The determining means determines values of the two or more components, respectively.
The comparison unit compares the value determined by the determination unit with a corresponding threshold value for each of the two or more components.
The specifying means specifies the pupil region of the human eye for each of the two or more components, and overlaps the pupil region specified for each of the two or more components, or for each of the two or more components The identification device according to claim 1, wherein a region obtained by combining the identified pupil regions is a pupil region of the human eye.   The identification device according to claim 4, wherein the two or more components include at least a red, green, or blue component.   The identification device according to claim 4, wherein the two or more components include at least a color difference component.   7. The specifying apparatus according to claim 1, further comprising a threshold value determination unit that determines the threshold value from a value of the designated component of a pixel in the region.   Threshold value determining means for determining a pixel corresponding to the pupil or the iris among the pixels of the region and determining the threshold value from the value of the designated component of the pixel corresponding to the pupil or the iris is further provided. The specifying device according to claim 1, wherein   9. The specifying apparatus according to claim 1, further comprising a processing unit that performs a process of increasing the resolution of the area detected by the detection unit. A detection step for detecting a region including human eyes in the image indicated by the image data;
For each pixel in the region, a determination step for determining a value of a designated component from information indicating the color of the pixel;
A comparison step for comparing the value determined in the determination step with a threshold for each pixel in the region;
A specifying step of specifying the pupil region of the human eye from the comparison result in the comparing step;
A method for specifying a pupil region, comprising:   A program that causes a computer to function as the specific device according to claim 1.

Images