JP2009239347A - Image processor and image processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor capable of exactly extracting a scene where a large majority of figures gaze at the same object of interest, among images including a plurality of figures as an object; and to provide an image processing program. <P>SOLUTION: The image processor comprises a face detection portion, an eye watching direction estimation portion, a distance calculation portion, and a cross determination portion. The face detection portion detects faces of the plurality of figures contained in an image signal based on image patterns in the image signal. The eye watching direction estimation portion estimates eye watching directions of the plurality of figures whose faces are detected by the face detection portion. The distance calculation portion calculates respective distances to the plurality of figures whose eye watching directions are estimated by the eye watching direction estimation portion. The cross determination portion determines whether the eye watching directions cross each other for the plurality of figures, using: the estimated result by the eye watching direction estimation portion; and the calculated result by the distance calculation portion. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing apparatus and an image processing program for performing predetermined processing on an image signal.

  2. Description of the Related Art Conventionally, in an image processing apparatus such as a digital camera having an imaging function, when a subject is a human, a technique for detecting the human face and focusing on the detected face position is known. In recent years, there is also known a technique for taking a picture by automatically recognizing the timing at which a human subject is showing an expression suitable for photographing such as a smile.

  Under such circumstances, a technique for extracting an image of a camera line of sight by determining the face direction and the position of the black eye of each person when photographing a plurality of persons is disclosed (for example, , See Patent Document 1). In addition, a technique is also disclosed in which a person's intention of action is determined by estimating a person's line-of-sight direction detected from a video and extracting the characteristic of the line-of-sight direction (see, for example, Patent Document 2).

JP 2007-96440 A JP 2007-6427 A

  When shooting with a plurality of persons as subjects, it is common to shoot in a state where all of them are looking at the camera, but there are other scenes suitable for shooting. For example, a scene in which the eyes of a plurality of people are gazing at the same object other than the image processing device can give a glimpse through the act of gazing at the same object in addition to creating a sense of unity in the image Because it can capture the moment of interaction between people, it is suitable for shooting.

  However, in the above-described prior art, it is not possible to extract a scene in which a plurality of persons are gazing at an object other than the image processing apparatus all at once. Specifically, in the technique described in Patent Document 1, only the camera line of sight of a plurality of persons is determined, and it is determined whether the lines of sight of the plurality of persons are concentrated on an object other than the camera. I couldn't. In addition, the technique described in Patent Document 2 merely estimates the face direction and line-of-sight direction of an individual person, and processing that considers the line-of-sight relationship among a plurality of persons in a video can be performed. There wasn't.

  For this reason, in order to extract a scene in which a plurality of persons who are subjects in the conventional image processing apparatus are gazing at an object other than the image processing apparatus all at once, the photographer himself has to make a determination by visual observation. It is difficult to extract such scenes when there are many subjects or when the subjects are far away.

  The present invention has been made in view of the above, and can accurately extract a scene in which a large number of persons are gazing at the same object in an image including a plurality of persons as subjects. An object of the present invention is to provide an image processing program.

  In order to solve the above-described problems and achieve the object, an image processing apparatus according to the present invention detects a plurality of human faces included in an image signal based on an image pattern in the image signal. A line of sight estimation unit that estimates the line of sight of the plurality of persons whose faces have been detected by the face detection unit, and a distance calculation unit that calculates distances to the plurality of persons from which the line of sight estimation unit has estimated the line of sight. And a cross determination unit that determines whether or not the eyes of the plurality of persons intersect by using the estimation result of the eye estimation unit and the calculation result of the distance calculation unit.

  The image processing apparatus according to the present invention further includes a display unit that displays image data corresponding to the image signal, and the display unit superimposes a result determined by the cross determination unit on the image data. It is characterized by displaying.

  In the image processing apparatus according to the present invention as set forth in the invention described above, the display unit superimposes and displays, on the image data, information that identifies a person whose eyes cross each other among the plurality of persons. It is characterized by that.

  In the image processing apparatus according to the present invention, the image processing apparatus further includes a trimming unit that trims the image data according to a determination result of the cross determination unit.

  In the image processing apparatus according to the present invention as set forth in the invention described above, the eye estimation unit is calculated by the face direction calculation unit that calculates the direction of the face detected by the face detection unit and the face direction calculation unit. And a line-of-sight direction calculation unit that calculates a direction in which a pupil included in the face is facing using the direction in which the face is facing.

  Further, in the image processing apparatus according to the present invention, in the above invention, the cross determination unit uses the person with the longest distance calculated by the distance calculation unit as a reference, and the line of sight of the reference person and the line of sight of another person. It is characterized by determining whether or not they intersect.

  The image processing apparatus according to the present invention is the image processing apparatus according to the above invention, wherein the cross determination unit is configured such that the line of sight of the reference person and the line of sight of the other person are far from the reference person with respect to the image processing apparatus. The case of crossing at is excluded from the object of determination.

  In the image processing apparatus according to the present invention, in the above invention, an imaging unit that captures an image of a subject within a predetermined visual field and generates an image signal, and imaging that instructs the imaging of the subject imaged by the imaging unit. An input unit to which an instruction signal is input, and a control unit that performs control for capturing an image of the subject being imaged by the imaging unit in accordance with the imaging instruction signal input from the input unit. To do.

  In the image processing apparatus according to the present invention as set forth in the invention described above, the control unit performs control to adjust exposure and / or focus to the face detected by the face detection unit.

  Further, in the image processing apparatus according to the present invention, in the above invention, the control unit performs control to automatically shoot a subject imaged by the imaging unit according to a result determined by the cross determination unit. It is characterized by.

  An image processing program according to the present invention provides a face for detecting a plurality of human faces included in an image signal based on an image pattern in the image signal in an image processing apparatus that performs predetermined processing on the image signal. Detecting step, estimating the eyes of the plurality of persons whose faces are detected in the face detecting step, storing the face information in a face information storage unit storing the face information, and determining the eyes in the eye estimation step The distance calculation step of calculating the distance to the person and storing it in the face information storage unit, the estimation result of the eye estimation step and the calculation result of the distance calculation step are read from the face information storage unit, and the read estimation result And a cross determination step of determining whether or not the eyes of the plurality of persons intersect by using the calculation result. And butterflies.

  According to the present invention, the eyes of a plurality of persons in an image signal are estimated, the distances to the persons who estimated the lines of sight are calculated, and the plurality of persons are calculated by using these eye line estimation results and distance calculation results. Therefore, it is possible to accurately extract a scene in which a large number of persons are gazing at the same object in an image including a plurality of persons as subjects.

  The best mode for carrying out the present invention (hereinafter referred to as “embodiment”) will be described below with reference to the accompanying drawings.

  FIG. 1 is a diagram schematically illustrating a configuration of an imaging apparatus 1 that is an image processing apparatus according to an embodiment of the present invention. The imaging device 1 captures an image of a subject within a predetermined visual field area and generates an image signal, a sensor unit 3 that detects the state of the visual field area of the imaging unit 2, and an image signal output by the imaging unit 2 An image processing unit 4 that performs image processing, a display unit 5 that displays information including image data output from the image processing unit 4, a display control unit 6 that controls display of the display unit 5, and compression and expansion of image data A compression / expansion unit 7, a recording medium interface 8 that reads information recorded on the recording medium 100 while recording information such as image data on a recording medium 100 such as a memory card that is externally attached to the imaging apparatus 1, and an imaging apparatus The operation of the input unit 9 to which one operation instruction signal is input, the audio output unit 10 that outputs audio of various types of information including warnings, the auxiliary light projection unit 11 that projects auxiliary light toward the imaging region, and the operation of the imaging device 1 control That the control unit 12, a storage unit 13 that stores various information about the operation of the imaging apparatus 1.

  The imaging unit 2 receives a photographing lens 21 that collects light from a subject in a predetermined field of view, a diaphragm 22 that adjusts an incident amount of light collected by the photographing lens 21, and incident light that has passed through the diaphragm 22. The image sensor 23 such as a CCD that converts the signal into an electrical signal, the analog signal processor 24 that performs various processes including sensitivity correction and white balance on the analog signal output from the image sensor 23, and the analog signal processor 24 output the signal. An A / D conversion unit 25 that generates digital image data from the signal is included. Among these, the photographing lens 21 is an optical system including one or a plurality of lenses, and is driven by the lens driving unit 14. The diaphragm 22 is driven by the diaphragm driving unit 15.

  The image processing unit 4 detects a person's face included in the image signal based on the image pattern of the image signal. When the face detection unit 41 detects a plurality of person's faces, The eye estimation unit 42 for estimating the distance, the distance calculation unit 43 for calculating the distance to the plurality of persons whose eye estimation has been performed by the eye estimation unit 42, the estimation result of the eye estimation unit 42, and the calculation result of the distance calculation unit 43 are used. Accordingly, a cross determination unit 44 that determines whether or not at least some of the eyes of a plurality of people intersect, and a trimming unit 45 that performs trimming of image data based on the determination result of the cross determination unit 44 Have. Of these, the eye line estimation unit 42 calculates the face direction calculation unit 421 that calculates the face direction (face direction) detected by the face detection unit 41 and the face direction calculated by the face direction calculation unit 421. And a gaze direction calculation unit 422 that calculates a direction in which the pupil included in the face is facing.

  The input unit 9 includes a power button of the imaging apparatus 1, a shutter button that gives an imaging instruction, a mode switching button that switches various operation motors included in the imaging apparatus 1, a control button that includes instructions for reproducing and editing image data, and the like. .

  The control unit 12 includes a clock 121 that determines the shooting date and time, and controls the operation of the imaging apparatus 1 according to the detection result of the sensor unit 3 and the operation instruction signal input from the input unit 9. The control unit 12 is realized using an MPU and is connected to each component to be controlled via a bus line.

  The storage unit 13 includes a semiconductor memory such as a flash memory or a RAM that is fixedly provided inside the imaging device 1, and includes a program storage unit 131 that stores various programs executed by the imaging device 1, and an image processing unit. And a face information storage unit 132 that stores face information such as the face direction and eye direction of the subject obtained in step 4.

  FIG. 2 is a diagram illustrating a part of operation modes that can be set in the imaging apparatus 1. In FIG. 2, the operation modes unique to the imaging device 1 are mainly described, and the description of the same operation mode that can be set by the conventional imaging device is omitted. When the imaging apparatus 1 is set to the shooting mode, a line-of-sight determination shooting mode in which shooting is performed after performing line-of-sight determination for determining a crossing state of a plurality of human eyes included in the image, or after normal shooting is performed. It is possible to further set an eye-gaze determination trimming mode for photographing that performs image trimming based on the eye-gaze determination and the eye-line determination result. Among these, when the imaging device 1 is set to the eye-gaze determination shooting mode, one or a plurality of modes among the trimming mode, the continuous shooting mode, and the automatic shooting mode can be further set. On the other hand, when the image pickup apparatus 1 is set to the reproduction mode, after performing eye-gaze determination when reproducing the image data recorded by the recording medium 100, reproduction eye-gaze determination for performing image trimming based on the determination result A trimming mode can be further set.

  FIG. 3 is a flowchart showing the flow of processing when the imaging apparatus 1 having the above configuration performs an operation including eye gaze determination processing.

  First, the case where the imaging device 1 is set to the shooting mode (step S1, shooting mode) will be described. When the imaging device 1 is set to the shooting mode, the display unit 5 displays the image data corresponding to the image signal output from the imaging unit 2 in real time under the control of the display control unit 6. Hereinafter, the image data displayed in real time on the display unit 5 is referred to as a “through image”. If the imaging device 1 is set to the eye-gaze determination shooting mode (step S2, Yes), the imaging device 1 performs eye-gaze determination processing on the through image (step S3).

  FIG. 4 is a flowchart illustrating an overview of eye-gaze determination processing performed by the imaging apparatus 1. First, the face detection unit 41 detects the face of a person present in the through image displayed on the display unit 5 (step S31). In performing this face detection, an existing method can be applied (for example, P. Viola and M. Jones, “Rapid Object Detection using a Boosted Cascade of Simple Features,” Proc. Of CVPR (2001). See). The control unit 12 performs control to focus on the position of the face detected by the face detection unit 41.

  When the face detection unit 41 detects a plurality of faces in the through image (step S32, Yes), the imaging device 1 proceeds to step S33. On the other hand, when the face detection unit 41 does not detect a plurality of faces in the through image (No at Step S32), the imaging device 1 proceeds to Step S4.

  In step S33, the eye line estimation unit 42 performs a line eye estimation loop process for estimating the eye line directions of the plurality of faces detected by the face detection unit 41 for each face (step S33). In this loop process, first, the face direction calculation unit 421 calculates the direction in which the face is facing (face direction) (step S331). The face direction calculation unit 421 obtains an image (face image) of the face portion detected by the face detection unit 41, and performs processing such as differential background and skin color region extraction on the obtained face image to determine the face region. After the extraction, the face direction is calculated using the extracted face area. FIG. 5 is a diagram illustrating an example of a face area extracted by the face direction calculation unit 421. The face area 31 shown in the figure is a rectangular area circumscribing the human head. The face direction calculation unit 421 that has extracted the face area 31 calculates the center position of the face area 31 (face area center position 32), and detects the position of the face organ such as the eyebrows, eyes, and nose, thereby detecting the face organ center. The position 33 is calculated. Thereafter, the face direction calculation unit 421 calculates the face direction on the horizontal plane and the face direction on the vertical plane by using the face region center position 32 and the face organ center position 33, respectively.

FIG. 6 is a diagram showing an outline of the face ellipse model used when calculating the face direction on the horizontal plane. The face 200 shown in the figure is an ellipsoid, and FIG. 6 shows a horizontal section passing through the center of the ellipsoid. The face 200 is directed horizontally right by an angle φ _{face with} respect to the optical axis direction (vertical direction in FIG. 6) of the photographing lens 21. The projection plane 300 corresponds to the face area 31. Further, the point O ′ _face corresponds to the face region center position 32 on the horizontal plane in FIG. 6, and the point C ′ _face corresponds to the face organ center position 33 on the horizontal plane in FIG. 6.

In FIG. 6, the face angle φ _face indicating the face direction on the horizontal plane of the face 200 is the horizontal width w _face of the face area 31 on the projection plane, the face area center position 32 (point O _face ) on the projection plane 300 and the face. By using the distance c _face to the organ center position 33 (point C) and the ratio k (face ellipse ratio) between the face width (= 2a) and the length in the depth direction of the face (= 2b) (1) ).

FIG. 7 is a diagram showing an outline of the face ellipse model used when calculating the face direction on the vertical plane. In FIG. 7, the vertical cross section which passes the center of the face 200 which is an ellipsoid is shown. The face 200 faces upward by an angle ψ _{face with} respect to the optical axis direction (left and right direction in FIG. 7) of the photographing lens 21. In FIG. 7, the face angle ψ _face indicating the face direction on the vertical plane of the face 200 is the vertical height h _face of the face area 31 on the projection plane 300 and the face area center position 32 (point O _{face on the} projection plane 300). ) and the distance c between the face organ center position 33 (point C) _'face, and the height of the face (= 2d) and the depth direction of the length of the face (= 2e) ratio k' of the following equation by using ( 2).

Thereafter, the line-of-sight direction calculation unit 422 calculates the line-of-sight direction using the face angle φ _face on the horizontal plane and the face angle ψ _face on the vertical plane calculated in step S331 (step S332). The eye direction calculation unit 422 detects a pupil from the face region 31 by a known method such as pattern matching, and detects an eye region from the detected pupil.

FIG. 8 is a diagram illustrating an outline of an eyeball model applied when the eye direction calculation unit 422 calculates the eye direction on a horizontal plane. Most of the spherical eyeball 400 shown in the figure is covered with skin, and the region actually appearing in the face image is an arc that reaches the point E ₂ ′ from the point E ₁ ′ through the pupil 401. The point O ′ _eye is the center of the eyeball 400, and the point O _eye is a point obtained by projecting the point O ′ _eye on the projection plane 500. Point I is a point obtained by projecting the center I ′ of the pupil 401 onto the projection plane 500. In FIG. 8, when the distance between the eyeball center position O _eye and the pupil center position I on the projection plane 500 is c _eye and the angle of the hidden portion of the eyeball 400 in the horizontal direction is α, the optical axis direction of the photographing lens 21 (FIG. The direction of the center I ′ of the pupil 401 (the vertical direction of 8), that is, the _eye line direction φ _eye on the horizontal plane is given by the following equation (3).

ここで、ｃ'_eyeは投影面５００における眼球中心位置Ｏ_eyeと瞳中心位置Ｉとの垂直方向の距離であり、角度α'は眼球４００の垂直方向の隠れ部分の角度である。 The eye direction calculation unit 422 applies the same eyeball model as that applied to the calculation of the eye direction on the horizontal plane when calculating the eye direction on the vertical plane. Therefore, the line-of-sight direction ψ _eye in the direction perpendicular to the optical axis direction of the photographing lens 21 is given by the following equation (4).

Here, c ′ _eye is the vertical distance between the eyeball center position O _eye and the pupil center position I on the projection plane 500, and the angle α ′ is the angle of the hidden part of the eyeball 400 in the vertical direction.

  When the line-of-sight direction calculation unit 422 can calculate the line-of-sight direction of both eyes, the line-of-sight direction is determined by weighting and summing the calculated line-of-sight directions of both eyes. Here, the weight for calculating the sum is determined according to the face direction. For example, when the face is directed to the right (left side as viewed from the image pickup apparatus 1), the right eye hardly enters the image, so the weight of the left eye is sufficiently larger than the weight of the right eye. When the face is facing the front, the average of the eye direction of both eyes is taken. The eye direction calculated by the eye direction calculation unit 422 in this way is stored in the face information storage unit 132 of the storage unit 13 as the eye estimation result.

  Note that the parameters for performing the eye direction calculation processing are not limited to those described above, and for example, the angle of the hidden part of the eyeball may be different between the center side and the outside of the face. The detailed contents of the face direction calculation process and the eye direction calculation process including this point are disclosed in, for example, Patent Document 2 described above.

  When the line-of-sight estimation unit 42 calculates a plurality of line-of-sight directions as a result of performing the line-of-sight estimation loop processing in step S33 on all faces (step S34, Yes), the imaging device 1 proceeds to step S35. On the other hand, when the line-of-sight estimation unit 42 does not calculate a plurality of line-of-sight directions as a result of performing the line-of-sight estimation loop processing in step S33 on all faces (step S34, No), the imaging device 1 The determination process ends and the process proceeds to step S4 described later.

FIG. 9 is a diagram schematically illustrating a situation in the vicinity of the imaging apparatus 1 that is referred to when explaining the eye-gaze determination process after step S35. Specifically, FIG. 9 shows a situation in which two persons A1 and A2 existing in the imaging region (horizontal angle of view θ _X ) of the imaging device 1 are paying attention to the same target T. It is the schematic diagram seen from the top. In FIG. 9, the position of the person A1 farther from the photographic lens 21 of the imaging device 1 is the coordinate origin O, and the axis parallel to the optical axis P of the photographic lens 21 is the L axis, while the light of the photographic lens 21 is The axis orthogonal to the axis P is taken as the X axis. In FIG. 9, the horizontal line-of-sight angles of the persons A1 and A2 (corresponding to _φeye in Expression (3)) are φ1 and φ2, respectively. In this sense, the angles φ1 and φ2 have been calculated in step S332 and stored in the face information storage unit 132.

  In step S <b> 35, the distance calculation unit 43 calculates the distances to the plurality of persons whose eye line directions are calculated starting from the photographing lens 21 (step S <b> 35). FIG. 10 is a diagram conceptually illustrating the calculation of the distances L1 and L2. For example, the distance calculation unit 43 calculates the distance L1 to the person A1 by focusing on the person A1, and then calculates the distance L2 to the person A2 based on the distance L1. In FIG. 10, the focal length of the taking lens 21 is F, the height of the face 201 of the person A1 and the height of the face 202 of the person A2 in the face image are H1 and H2, respectively. Both are close to D. Using these definitions, the distances L1 and L2 from the taking lens 21 to the persons A1 and A2 are expressed as L1 = DF / H1 and L2 = DF / H2, respectively. Therefore, the distance calculation unit 43 can calculate the distance L2 = (H2 / H1) L1 by using the distance L1. The distances L1 and L2 calculated by the distance calculation unit 43 are stored in the face information storage unit 132 of the storage unit 13.

Thereafter, the cross determination unit 44 determines whether or not the eyes of a plurality of people intersect (step S36). First, the cross determination unit 44 reads the distances L1 and L2 calculated by the distance calculation unit 43 in step S35 from the face information storage unit 132, and calculates the L coordinate difference ΔL and the X coordinate difference ΔX between the person A1 and the person A2. It calculates according to following Formula (5) and (6), respectively.
ΔL = L1-L2 (5)
ΔX = L1 tan (ρ1) + L2 tan (ρ2) (6)
Here, ρ1 (<θ _X ) in Expression (6) is an angle formed by the direction of the person A1 viewed from the photographing lens 21 and the optical axis P of the photographing lens 21. Further, ρ2 (<θ _X ) in Expression (6) is an angle formed by the direction of the person A2 viewed from the photographing lens 21 and the optical axis P of the photographing lens 21.

FIG. 11 is a schematic view of the same situation as FIG. 9 as viewed from the side of the imaging apparatus 1. In FIG. 11, as in FIG. 9, the position of the person A1 with the larger distance from the imaging device 1 is the coordinate origin O, and the axis orthogonal to the optical axis P of the photographic lens 21 is the Y axis. In FIG. 11, the angle of view of the imaging apparatus 1 in the Y-axis direction is θ _Y. Further, in FIG. 11, (corresponding to [psi _eye of formula (4)) horizontal eye direction angles of the person A1, A2 respectively .psi.1, have a .psi.2. In this sense, the angles ψ 1 and ψ 2 have already been calculated in step S 332 in the same manner as φ 1 and stored in the face information storage unit 132. The cross determination unit 44 calculates the difference ΔY between the Y coordinates of the persons A1 and A2 according to the following equation (7).
ΔY = L1 tan (η1) + L2 tan (η2) (7)
Here, η1 (<θ _Y ) in Expression (7) is an angle formed by the direction of the person A1 viewed from the photographing lens 21 and the optical axis P of the photographing lens 21. In addition, η2 (<θ _Y ) in Expression (7) is an angle formed by the direction of the person A2 viewed from the photographing lens 21 and the optical axis P of the photographing lens 21.

Subsequently, the cross determination unit 44 uses the horizontal line-of-sight angles φ1 and φ2 of the persons A1 and A2 read from the face information storage unit 132 in addition to the calculation results of the expressions (5) and (6). The intersection coordinates (Ls, Xs) of the eyes of the persons A1 and A2 in the horizontal direction are calculated according to the following equations (8) and (9), respectively.

Further, the cross determination unit 44 uses, in addition to the calculation results of the expressions (5) and (7), the angles ψ1 and ψ2 in the vertical line-of-sight direction of the persons A1 and A2 read from the face information storage unit 132, The intersection coordinates (Ls, Ys) of the eyes of the persons A1 and A2 in the vertical direction are calculated according to the following equations (10) and (11).

  Next, the cross determination unit 44 determines whether Ls in Expression (8) matches Ls in Expression (10). Here, “match” means that the difference between the two Ls is a value within a predetermined range in view of the fact that the two Ls are estimated values. When Ls in Expression (8) matches Ls in Expression (10), the cross determination unit 44 determines that the line of sight of the person A1 and the line of sight of the person A2 are concentrated.

  In addition, when Ls becomes ∞ in Expressions (8) and (10), the cross determination unit 44 determines that the eye direction is parallel and intersects at infinity. In addition, when Ls <0 in Expressions (8) and (10), the eyes of the two people intersect on the back side of the face, so the cross determination unit 44 excludes this case from the determination target.

  Here, a case where three or more persons are included in the image data will be described. In this case, the cross determination unit 44 uses the person located farthest from the photographing lens 21 as a reference person, and performs cross determination using a difference in coordinate values between the reference person and another person. For example, when n (n is a natural number of 3 or more) persons are included in the through image, the cross determination unit 44 sets the intersection coordinates between the line-of-sight direction of the reference person and the line-of-sight direction of another person to the maximum (n− 1) Calculate as follows. Subsequently, the cross determination unit 44 determines whether or not the horizontal direction Ls (Equation (8)) and the vertical direction Ls (Equation (10)) match each of the calculated plurality of intersection coordinates. To do. As a result of this determination, if there are a plurality of intersection coordinates (hereinafter referred to as “normal intersection coordinates”) in which the horizontal direction Ls and the vertical direction Ls match, the cross determination unit 44 determines whether the plurality of normal intersection coordinates match. Determine whether or not. When performing this determination, the cross determination unit 44 extracts one normal intersection coordinate (reference intersection coordinate) to be used as a reference according to a predetermined rule, and calculates the difference between this reference intersection coordinate and the other normal intersection coordinates as components. Each is calculated. The cross determination unit 44 determines that the reference intersection coordinates and the normal intersection coordinates in which the difference for each component is within a predetermined range are the same. When the normal intersection coordinates that coincide with the reference intersection coordinates exist at a predetermined ratio or more, the cross determination unit 44 determines that the eyes of the person in the through image are concentrated.

  The process following the eye line determination process in step S3 described above differs depending on whether or not the imaging apparatus 1 is set to the automatic shooting mode. First, the case where the imaging device 1 is not set to the automatic shooting mode (step S4, No) will be described. In this case, when it is determined in step S3 that the eyes of a plurality of persons in the through image are concentrated (step S5, Yes), the display unit 5 determines that the eyes are concentrated as a determination result of the cross determination unit 44. Then, an OK display for notifying that the timing is suitable for shooting is displayed (step S6). On the other hand, if it is not determined that the eyes of a plurality of persons in the through image are concentrated (No in step S5), the imaging apparatus 1 does not output an OK display on the display unit 5 and proceeds to step S7 described later. Proceed directly.

  FIG. 12 is a diagram schematically illustrating an example of display output of OK display on the display unit 5. A display image 51 shown in the figure shows a display example under the situation shown in FIGS. In the display image 51, the character string “timing OK” is displayed superimposed on the upper left portion of the through image. When the OK display is displayed and output on the display unit 5, the photographer can recognize that the eyes of the persons A1 and A2 are concentrated and that the timing is suitable for shooting. In FIG. 12, the eyes (indicated by broken lines) of the persons A1 and A2 and the intersection CP of the two eyes are merely auxiliary.

  FIG. 13 is a diagram illustrating another display output example (second example) of OK display. In the display image 52 shown in the figure, a square-shaped eye-focused person identification symbol 52a indicating a person whose eyes are concentrated is displayed superimposed on the face areas of the persons A11 and A12 in the through image, and the eyes are displayed. A triangular eye-unconcentrated person identification symbol 52b facing in a different direction from other persons is displayed superimposed on the face area of the person A13 in the through image. The gaze-concentrated person identification symbol 52a and the gaze-unconcentrated person identification symbol 52b are displayed when there are more than a predetermined percentage of persons with concentrated eyes in the through image. Note that the display unit 5 may display the gaze-concentrated person identification symbol 52a and the gaze-unconcentrated person identification symbol 52b in different colors. In addition, the condition (ratio of the person who is focused on the whole subject) for the eyes-focused person specifying symbol 52a and the eyes non-focused person specifying symbol 52b to be displayed superimposed on the through image is as follows. 9 may be used to arbitrarily input settings.

  FIG. 14 is a diagram showing still another display output example (third example) of OK display. A display image 53 shown in the figure displays the results of eye line determination of five persons A21 to A25. In the display image 53, a message indicating that there are four of five persons whose eyes are concentrated is superimposed on the through image, and the person A25 whose eyes are different from the other four is displayed. A non-concentrated person identification symbol 53a is displayed superimposed on the through image in the face area. Such messages and symbols are displayed when there are more than a predetermined percentage of people whose eyes are concentrated in the through image.

  Thereafter, when a shooting instruction signal is input from the input unit 9 by the photographer (step S7, Yes), the imaging apparatus 1 captures an image under the control of the control unit 12 (step S8) and takes the image. The image data is compressed to a predetermined size by the compression / decompression unit 7, and the compressed image data is recorded on the recording medium 100 by the recording medium interface 8 (step S9). Here, the imaging apparatus 1 can also write and record the captured image data in the storage unit 13. Note that if the shooting instruction signal is not input even after a predetermined time has elapsed in step S7 (step S7, No), the imaging apparatus 1 returns to step S3.

  Next, the case where the imaging device 1 is set to the automatic shooting mode will be described (step S4, Yes). In this case, when it is determined that the lines of sight of a plurality of persons in the through image are concentrated as a result of the line-of-sight determination process (step S3) by the image processing unit 4 (step S10, Yes), the imaging apparatus 1 automatically Then, the photographing process is performed (step S8), the photographed image data is compressed to a predetermined size by the compression / decompression unit 7, and the compressed image data is recorded on the recording medium 100 by the recording medium interface 8 (step S9). On the other hand, when it is determined that the eyes of a plurality of persons in the through image are not concentrated as a result of the eye line determination process by the image processing unit 4 (No in step S10), the imaging device 1 returns to step S3.

  The processing following step S9 differs depending on whether or not the continuous shooting mode is set in the imaging apparatus 1. When the imaging device 1 is not set to the continuous shooting mode (No at Step S11), the image processing unit 4 reads out the image data from the recording medium interface 8 and performs the same eye line determination processing (Step S12) as Step S3 described above. Do. The reason why the line of sight determination process is performed again is that there is a time lag until the actual shooting is performed after the line of sight determination process in step S3, and a determination result different from the determination result in step S3 may occur.

  On the other hand, when the imaging device 1 is set to the continuous shooting mode (step S11, Yes), if shooting for the predetermined number of shots has not been completed (step S13, No), the process returns to step S8 and the shooting is repeated. Do. If the imaging device 1 is set to the continuous shooting mode (step S11, Yes) and shooting for a predetermined number of shots has been completed (step S13, Yes), the image processing unit 4 starts from the recording medium interface 8. The image data is read out, and the eye line determination process similar to that in step S3 is performed (step S12).

  If the image processing unit 4 determines that the eyes of the person included in the captured image data are concentrated as a result of performing the line-of-sight determination process (step S12) (step S14, Yes), the imaging device 1 displays the image. Data is recorded on the recording medium 100 through the recording medium interface 8 as a focused eye image (step S15). The “line of sight image” here is image data to which information indicating that the line of sight is concentrated is given. For example, a flag indicating that the image is a line of sight image with respect to the original image data. It is a thing that stands. Note that when the image processing unit 4 does not determine that the human eyes in the captured image data are concentrated as a result of the eye determination processing (No in step S14), the imaging device 1 returns to step S3. .

  Thereafter, when the imaging apparatus 1 is set to the trimming mode (step S16, Yes), the trimming unit 45 performs trimming of the eye-gaze concentrated image (step S17). Subsequently, the display unit 5 displays the result trimmed by the trimming unit 45 (step S18). FIG. 15 is a diagram illustrating a display example of the trimming result on the display unit 5. In the trimming result display image 54 shown in the figure, an area where only the persons A11 and A12 whose eyes are concentrated is extracted is displayed as a trimming area 54a using a broken line. Note that instead of the display unit 5 displaying the trimming result display image 54 in step S18, only the trimming region 54a may be displayed. Further, a message asking whether or not to record the trimmed image may be output at an appropriate position of the trimming result display image 54 to prompt the photographer to make a selection input from the input unit 9.

  When an operation instruction signal for recording a trimmed image is input from the input unit 9 (step S19, Yes), the image pickup apparatus 1 compresses the trimmed image data to a predetermined size by the compression / expansion unit 7, and the compressed trimming is performed. Image data is recorded on the recording medium 100 by the recording medium interface 8 (step S20). On the other hand, when the operation instruction signal for recording the trimmed image is not input (step S19, No), the imaging apparatus 1 ends the series of processes.

  When the imaging device 1 is not set to the trimming mode (No at Step S16), the imaging device 1 ends a series of processes.

  Next, the case where the imaging apparatus 1 is set to the shooting mode (step S1, shooting mode) and is not set to the eye-gaze determination shooting mode (step S2, No) will be described. In this case, the imaging apparatus 1 performs a normal shooting process (step S21).

  The processing following the normal shooting processing differs depending on whether or not the imaging apparatus 1 is set to the shooting eye determination trimming mode. When the imaging device 1 is set to the imaging eye-gaze determination trimming mode (step S22, Yes), the image processing unit 4 performs the same eye-line determination processing as that of step S12 on the captured image data (step S23). On the other hand, when the imaging apparatus 1 is not set to the imaging eye-gaze determination trimming mode (step S22, No), the imaging apparatus 1 performs only normal imaging processing.

  Hereinafter, a case where the imaging device 1 is set to the photographing eye-line determination trimming mode will be described. In this case, when the image processing unit 4 determines that the eyes of the person included in the image data are concentrated as a result of performing the line-of-sight determination process (step S23) (step S24, Yes), the imaging device 1 The image data is recorded on the recording medium 100 through the recording medium interface 8 as a focused eye image (step S25). Subsequently, the trimming unit 45 trims the eye-growth concentrated image (step S17). The processes in steps S18 to S20 subsequent to step S17 are as described above.

  On the other hand, when the image processing unit 4 does not determine that the human eyes included in the image data are concentrated as a result of performing the line-of-sight determination process (step S24, No), the imaging device 1 performs a series of processes. finish.

  Next, the case where the reproduction mode is set in the imaging device 1 (step S1, reproduction mode) will be described. First, the display unit 5 reproduces and displays the image data selected by the photographer from the image data stored in the recording medium 100 (step S26).

  When the imaging apparatus 1 is set to the reproduction eye-gaze determination trimming mode (step S27, Yes), the image processing unit 4 performs eye-gaze determination processing on the image data being reproduced and displayed (step S23). The processing after step S24 following step S23 is the same as when the imaging mode and the eye-gaze determination trimming mode for imaging are set in the imaging apparatus 1.

  On the other hand, when the reproduction eye determination trimming mode is not set in the imaging device 1 (step S27, No), the imaging device 1 ends the process.

  According to the embodiment of the present invention described above, the lines of sight of a plurality of persons in an image signal are estimated, the distances to the persons who estimated the lines of sight are calculated, and the line of sight estimation results and distance calculation are calculated. In order to determine whether or not the eyes of multiple persons intersect by using the results, it is possible to accurately extract the scene where the majority of people are gazing at the same object in an image that includes multiple persons as subjects. It becomes possible to do.

  The best mode for carrying out the present invention has been described so far, but the present invention should not be limited only by the above-described embodiment. For example, when the photographer moves the imaging apparatus 1 and an intersection of the eyes of a plurality of persons enters a through image displayed on the display unit 5, the intersection is displayed on the display unit 5. Also good. In FIG. 16, as a result of the photographer moving the imaging apparatus 1 in the positive direction of the L axis from the state shown in FIG. 9 (see the display image 51 in FIG. 12), the target T enters the visual field region of the imaging apparatus 1. It is a figure which shows the example of a display in the display part 5 in the case. In the display image 55 shown in FIG. 16, the intersection CP of the line of sight of the person A1 and the line of sight of the person A2 is displayed by an asterisk. For this reason, the photographer can easily recognize the position where the lines of sight of a plurality of persons intersect and the object existing at the intersection position.

  Thus, the present invention can include various embodiments and the like not described herein, and various design changes and the like can be made without departing from the technical idea specified by the claims. It is possible to apply.

It is a figure which shows the structure of the image processing apparatus (imaging device) which concerns on one embodiment of this invention. It is a figure which shows the example of the processing mode in the image processing apparatus (imaging apparatus) which concerns on one embodiment of this invention. It is a flowchart which shows the flow of a process including the eyes | visual_axis determination process of the image processing apparatus (imaging device) which concerns on one embodiment of this invention. It is a flowchart which shows the outline | summary of a gaze determination process. It is a figure which shows the structure of a face area | region. It is a figure which shows the outline | summary of a face ellipse model (horizontal direction). It is a figure which shows the outline | summary of a face ellipse model (vertical direction). It is a figure which shows the outline | summary of an eyeball model. It is the schematic diagram which looked at the relationship between the several person who is a subject, and the target which these several persons are paying attention from the upper part of the image processing apparatus. It is a figure which shows the outline | summary of the distance calculation to a to-be-photographed object. It is the schematic diagram which looked at the relationship between the several person who is a subject and the target which these several persons are paying attention from the side of the image processing apparatus. It is a figure which shows the example of a display (1st example) of OK display. It is a figure which shows the example of a display (2nd example) of OK display. It is a figure which shows the example of a display of OK display (3rd example). It is a figure which shows the example of a display of a trimming result. It is a figure which shows the example of a display in the display part after the eyes | visual_axis determination process of the image processing apparatus (imaging apparatus) which concerns on other embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Imaging device 2 Imaging part 3 Sensor part 4 Image processing part 5 Display part 6 Display control part 7 Compression / decompression part 8 Recording medium interface 9 Input part 10 Audio | voice output part 11 Auxiliary light projection part 12 Control part 13 Storage part 14 Lens drive part DESCRIPTION OF SYMBOLS 15 Aperture drive part 21 Shooting lens 22 Aperture 23 Imaging element 24 Analog signal processing part 25 A / D conversion part 31 Face area 32 Face area center position 33 Face organ center position 41 Face detection part 42 Eye estimation part 43 Distance calculation part 44 Cross Determination unit 45 Trimming unit 51, 52, 53, 55 Display image 52a Eye-concentrated person specifying symbol 52b, 53a Eye-inconcentrated person specifying symbol 54 Trimming result display image 54a Trimming area 100 Recording medium 121 Clock 131 Program storage unit 132 Face information storage Part 200, 201, 202 Face 300 , 500 Projection plane 400 Eyeball 401 Pupil 421 Face direction calculation unit 422 Eye direction calculation unit A1, A2, A11 to A13, A21 to A25 Person CP Intersection T Target

Claims (11)

A face detection unit for detecting faces of a plurality of persons included in the image signal based on an image pattern in the image signal;
A line of sight estimation unit for estimating the lines of sight of the plurality of persons whose faces have been detected by the face detection unit;
A distance calculator that calculates distances to the plurality of persons whose eyes are estimated by the eyes estimator;
A cross determination unit that determines whether or not the eyes of the plurality of persons intersect by using the estimation result of the eye estimation unit and the calculation result of the distance calculation unit;
An image processing apparatus comprising: A display unit for displaying image data corresponding to the image signal;
The display unit
The image processing apparatus according to claim 1, wherein a result determined by the cross determination unit is displayed superimposed on the image data. The display unit
The image processing apparatus according to claim 2, wherein information for identifying a person whose eyes cross each other among the plurality of persons is displayed superimposed on the image data.   The image processing apparatus according to claim 1, further comprising a trimming unit that trims the image data according to a determination result of the cross determination unit. The eye estimation unit
A face direction calculation unit that calculates a direction of the face detected by the face detection unit;
A line-of-sight direction calculation unit that calculates a direction in which a pupil included in the face is facing using the direction in which the face is calculated, which is calculated by the face direction calculation unit;
The image processing apparatus according to claim 1, further comprising: The cross determination unit
6. The method according to claim 1, further comprising: determining whether or not the line of sight of the reference person and the line of sight of another person intersect based on the person having the longest distance calculated by the distance calculation unit. An image processing apparatus according to claim 1. The cross determination unit
The case where the line of sight of the reference person and the line of sight of the other person intersect with the image processing apparatus at a position farther than the reference person is excluded from determination targets. The image processing apparatus described. An imaging unit that captures an image of a subject within a predetermined visual field and generates an image signal;
An input unit to which a shooting instruction signal for instructing shooting of a subject imaged by the imaging unit is input;
A control unit that performs control to capture a subject imaged by the imaging unit in response to the imaging instruction signal input from the input unit;
The image processing apparatus according to claim 1, further comprising: The controller is
The image processing apparatus according to claim 8, wherein control is performed to adjust exposure and / or focus to the face detected by the face detection unit. The controller is
10. The image processing apparatus according to claim 8, wherein control is performed to automatically photograph the subject being imaged by the imaging unit according to a result determined by the cross determination unit. In an image processing apparatus that performs predetermined processing on an image signal,
A face detection step of detecting faces of a plurality of persons included in the image signal based on an image pattern in the image signal;
Eye estimation step for estimating the eyes of the plurality of persons whose faces have been detected in the face detection step and storing them in a face information storage unit for storing face information;
A distance calculating step of calculating distances to the plurality of persons whose eyes are estimated in the eyes estimating step and storing them in the face information storage unit,
The estimation result of the eye estimation step and the calculation result of the distance calculation step are read from the face information storage unit, and by using the read estimation result and the calculation result, it is determined whether or not the eyes of the plurality of persons intersect. A cross determination step for determining,
An image processing program for executing

Images