JP4883013B2 - Face image processing device - Google Patents

Description

  The present invention relates to a face image processing apparatus that acquires a face image of a driver of a vehicle and performs image processing based on the acquired face image.

2. Description of the Related Art Conventionally, a technique for acquiring a driver's face image of a vehicle and performing image processing based on the acquired face image to detect the driver's eyes is known (for example, see Patent Document 1). In such a technique, when the driver wears glasses, the edge image reflected on the lens may be erroneously detected as an eye due to the reflection on the lens of the glasses. Therefore, the detection accuracy of the eyes is improved by irradiating the driver with light and removing the reflection image in which the reflection position moves as a reflection image of an object having a glossy reflection surface.
JP 2002-352229 A

  However, in the above prior art, since the reflection image whose reflection position moves is excluded from the eye candidates, for example, an eyelid or the like to be detected originally is excluded from the eye candidates, and there is a possibility that the eye detection accuracy is lowered.

  In addition, as a technique for detecting the eye by removing the image reflected in the glasses, the near-infrared LED (light emitting diode) is used to switch the illumination ON / OFF, and the image difference processing is performed. There is something that removes the reflected image. However, this technique requires a dedicated device for removing the reflected image, and there is a problem that the cost increases and a space for installing a new device needs to be secured. In addition, there is a problem that calculation load increases due to image difference processing or the like.

  In addition, as another technique, LED lighting is installed in the front direction of the driver, red-eye reflection is intentionally generated on the face image, and the red-eye component is extracted preferentially, so that the image reflected in the glasses is displayed. Some reduce the impact. However, this technique has a problem that a new device is necessary and a calculation load increase as in the above technique.

  Further, as another technique, there is a technique that learns image data of an image reflected in glasses and removes an image reflected in glasses based on the learned image data. However, with this technique, there is a problem that it is necessary to newly store dedicated learning data, a problem that a new dedicated algorithm for removing the reflected image is necessary, a calculation load increases, and processing time increases. There is a problem of increasing.

  The present invention has been made to solve such a problem, and provides a face image processing apparatus capable of detecting an image reflected in glasses with a simple configuration while suppressing an increase in calculation load. The purpose is to do.

  A face image processing apparatus according to the present invention is an image processing apparatus including an image processing unit that acquires a face image of a driver of a vehicle, performs image processing based on the acquired face image, and detects a driver's eyes. Based on the information on the movement of the vehicle acquired by the movement information acquisition means, the movement information acquisition means for acquiring the information on the movement of the vehicle, and the movement state of the edge image reflected on the glasses lens attached to the driver is estimated And a reflection detection means for detecting a reflection edge image that is an edge image reflected on the eyeglass lens based on the movement state of the edge image estimated by the estimation means. . Further, the image processing means does not use the reflection edge image detected by the reflection detection means as the eye candidate when detecting the eyes.

  According to such a face image processing device, the information about the movement of the vehicle is acquired, and based on the acquired information, the movement state of the edge image reflected in the driver's eyeglass lens is estimated, and the estimated edge image The edge image actually reflected on the eyeglass lens is detected on the basis of the movement state of. Since the edge image resulting from the scenery outside the vehicle moves with the moving state of the vehicle, the moving state of the edge image reflected on the eyeglass lens can be estimated based on the moving state of the vehicle. As a result, an edge image whose position changes in the same manner as the estimated movement state can be excluded as an unnecessary edge from candidates for eye detection (hereinafter referred to as “eye candidates”), and the eye detection accuracy Can be improved. Further, an unnecessary edge image reflected on the eyeglass lens can be detected and removed with a simple configuration without newly adding a dedicated device. The “eyeglass lens” includes a transparent body and a translucent body disposed in front of the eye, such as a lens for glasses and a glass for sunglasses. Further, the “edge image reflected on the eyeglass lens” includes an image caused by a linear portion (such as an end) of the object projected on the eyeglass lens. The detection of “eyes” includes recognition of the upper eyelid, the lower eyelid, and the pupil.

  Moreover, it is preferable that the moving state of the edge image includes the moving direction of the edge image. Accordingly, an edge image that moves in the same direction as the estimated movement direction of the edge image can be set as an unnecessary edge image reflected on the eyeglass lens and excluded from the eye candidates.

  Moreover, it is preferable that the movement state of the edge image includes an amount of movement of the edge image per unit time. As a result, an edge image that moves in the same manner as the estimated movement direction of the edge image and the amount of movement per unit time can be set as an unnecessary edge image reflected on the eyeglass lens and excluded from the eye candidates. .

  Further, it is preferable that the reflection detection means does not detect an edge image moving downward as a reflection edge image. When the vehicle moves forward, the reflected edge image moves upward on the eyeglass lens in accordance with the moving direction of the vehicle, so the edge image that moves downward can be an edge image reflected on the eyeglass lens. The nature is low, and conversely, there is a high possibility that the eye (upper eyelid, lower eyelid). For this reason, eye detection accuracy is improved by not setting the edge image that moves downward as the edge image reflected on the eyeglass lens.

  Further, it is preferable that the reflection detection means does not detect an edge image that moves in the vertical direction within a predetermined time as a reflection edge image. When the vehicle moves forward, the reflected edge image moves upward on the eyeglass lens in accordance with the moving direction of the vehicle, so the edge image that moves up and down within a predetermined time is reflected on the eyeglass lens. The possibility of being an edge image is extremely low, and conversely, the possibility of being an eye is high. Therefore, the eye detection accuracy is improved by not setting the edge image that moves up and down within a predetermined time as the edge image reflected on the eyeglass lens.

  According to the present invention, it is possible to provide a face image processing apparatus capable of detecting an image reflected in glasses with a simple configuration while suppressing an increase in calculation load.

  Preferred embodiments of the present invention will be described below with reference to the drawings. In the description of the drawings, the same or corresponding elements are denoted by the same reference numerals, and redundant description is omitted. FIG. 1 is a configuration diagram illustrating a face image processing apparatus according to an embodiment of the present invention.

  A face image processing apparatus 10 shown in FIG. 1 is mounted on a vehicle (hereinafter referred to as “own vehicle”) and acquires a face image of the driver D. The face image processing apparatus 10 includes a face image capturing camera 12 that captures a face image of the driver D, and a face image processing electronic control unit (hereinafter, “ 14 ”(referred to as“ face image processing ECU ”).

  The face image capturing camera 12 is installed on the upper surface of the column cover 16, for example, and acquires the face image of the driver D. The face image capturing camera 12 is, for example, a near-infrared camera, and receives near-infrared light irradiated from a near-infrared projection device (not shown) and receives reflected light reflected from an object (the face of the driver D), thereby driving. The face of the person D is imaged. For example, the face image capturing camera 12 can acquire an image of 30 frames per second. Then, the image signal regarding the face image of the driver D acquired by the face image capturing camera 12 is transmitted to the face image processing ECU 14.

  The face image processing ECU 14 includes a CPU that performs arithmetic processing, a ROM and a RAM that serve as a storage unit, an input signal circuit, an output signal circuit, a power supply circuit, and the like. The face image processing ECU 14 performs image processing on the input image signal and performs the face of the driver D Detect images. The detected face image is used for, for example, eye opening degree determination (awakening degree determination), aside look determination of the driver D and the like.

  Here, the face image processing apparatus 10 of the present invention acquires information related to the movement of the host vehicle, and detects a reflected edge image reflected on the eyeglass lens of the driver D based on the acquired information. And a function of excluding unnecessary reflection edge images from the eye candidates.

  The face image processing ECU 14 includes a vehicle speed sensor 18 that detects the host vehicle speed, a winker sensor 20 that detects a direction indicated by a winker (direction indicator), and a steering sensor 22 that detects a steering direction and a steering angle when the steering operation is performed. Connected. The vehicle speed sensor 18, the blinker sensor 20, and the steering sensor 22 function as movement information acquisition means of the present invention that acquires information related to the movement of the host vehicle.

  The CPU of the face image processing ECU 14 executes a program stored in the storage unit, whereby a face feature point extraction unit 24, a face orientation angle detection unit 26, an edge movement direction estimation unit 28, a reflection edge detection unit 30, An eye detection unit 32 is configured.

  The storage unit of the face image processing ECU 14 stores a plurality of three-dimensional face models in addition to a program for operating the CPU. The storage unit functions as a storage unit that stores captured image data and data related to facial feature points. In addition, data related to the eyeglass lens of the driver D (for example, the shape, size, curvature, etc. of the lens) may be stored in advance in the storage unit.

  The face feature point extraction unit 24 performs image processing based on the input image signal, recognizes the nose and mouth, and uses, for example, an edge-based method, a pattern recognition method, etc. The position (axis) Y is detected. For example, the face center position Y is detected based on the positions of the nose and mouth. Further, the face feature point extraction unit 24 performs edge processing based on the image around the eyes and detects a horizontal edge image around the eyes. The “horizontal edge image” is a linear image extending in the lateral direction of the face, and is a candidate for eyes (upper eyelid, lower eyelid).

  The face angle detection unit 26 detects the face angle of the driver D. Here, the face orientation angle is calculated from the relative positional relationship between the face width X detected by the face feature point extraction unit 24 and the face center position Y. The calculation of the face orientation angle is an existing technique, and the face orientation angle may be detected using other methods. As the “face-facing angle”, an angle with respect to the reference is given with the driver D facing the front of the vehicle as a reference.

  The edge movement direction estimation unit 28 determines the movement state of the host vehicle based on the host vehicle speed input from the vehicle speed sensor 18, the indication direction input from the turn signal sensor 20, the steering direction and the steering angle input from the steering sensor 22. The movement state of the reflection edge image (horizontal edge image) reflected on the spectacle lens is estimated based on the movement state of the host vehicle and the face orientation angle. Here, the moving direction and moving speed of the reflected edge image reflected on the spectacle lens are estimated. Examples of the “movement state” include a movement direction, a movement speed, and a movement amount.

  FIG. 2 is a schematic diagram showing a temporal change of an image around the eye, and shows a positional change of a reflected edge image reflected on the spectacle lens. As shown in FIG. 2, an edge image E <b> 1 extending in the horizontal direction (lateral direction of the face) can be recognized on the eyeglass lens 40. FIG. 2 shows the passage of time from the top (A) to the bottom (C). In the case shown in FIG. 2, the edge image E1 moves upward on the eyeglass lens 40 with the passage of time. ing.

  For example, when the host vehicle is moving forward, the reflection edge image reflected on the eyeglass lens 40 due to the scenery outside the vehicle moves upward on the eyeglass lens 40. Images are acquired at regular time intervals (for example, 30 frames per second), and the amount of movement (moving speed) per unit time of the edge image E1 can be determined by comparing consecutive frames.

  The reflection edge detection unit 30 searches for a horizontal edge image on the eyeglass lens 40, and an edge image E1 that matches the movement direction and movement speed calculated by the edge movement direction estimation unit 28 is reflected on the eyeglass lens 40. Detected as a reflection edge image.

  The reflection edge detection unit 30 does not detect the horizontal edge image as a reflection edge image when the detected horizontal edge image moves downward. This is because an edge image that moves downward is likely to be an upper eyelid or an edge image caused by the lower eyelid, so that an object that moves downward on the eyeglass lens 40 is not detected as a reflected edge image. I have to.

  The reflection edge detection unit 30 does not detect the horizontal edge image as a reflection edge image when the detected horizontal edge image moves up and down within a predetermined time. This is because the edge image that moves in the vertical direction per unit time is likely to be an edge image due to the upper eyelid or the lower eyelid, so that what moves downward on the eyeglass lens 40 is a reflection edge. The image is not detected.

  The eye detection unit 32 excludes the reflection edge detected by the reflection edge detection unit 30 from the eye candidates, and performs eye detection processing for detecting, from the eye candidates, those having a high fitness value (probable). And recognize the eyes of the driver D (upper eyelid, lower eyelid, pupil, etc.). For example, by referring to a three-dimensional face model stored in the storage unit, a likely eye candidate is recognized as an eye.

  Next, the operation of the face image processing apparatus 10 configured as described above will be described. FIG. 3 is a flowchart showing a processing procedure executed by the face image processing ECU according to the first embodiment of the present invention. First, the face image processing ECU 14 receives an image signal from the face image capturing camera 12, and detects the facial features of the driver D (S1). Here, the face image processing ECU 14 detects, for example, the width X of the face of the driver D and the center position Y of the face from the positions of the nose and mouth using existing technology.

  Next, the face image processing ECU 14 calculates the face angle of the driver D based on the detected relative position of the width X of the face of the driver D and the center position Y of the face (S2). Subsequently, the face image processing ECU 14 detects the host vehicle speed based on the signal from the vehicle speed sensor 18, detects the course change of the host vehicle, and turns right and left based on the signal from the blinker sensor 20, and the steering sensor 22. The steering direction and steering angle of the host vehicle are detected based on the signal from (S3).

  Next, the face image processing ECU 14 determines the flow direction of the horizontal edge image on the eyeglass lens based on the detected face orientation angle, the own vehicle speed, the course change and right / left turn of the own vehicle, the steering direction and the steering angle of the own vehicle. (Moving direction) and the moving amount are estimated for each of the left and right lenses (S4). For example, it may be estimated how many times a series of reflection edge images can be detected on a spectacle lens.

  Subsequently, the face image processing ECU 14 sets a search range around the eyes based on the recognized position of the nose and mouth, and detects a horizontal edge image around the eyes using a method such as a sobel filter (for example). S5). At this time, an edge image caused by the actual eye, an edge image reflected on the eyeglass lens due to the scenery outside the vehicle, and the like are detected as horizontal edges.

  Next, the face image processing ECU 14 searches the detected horizontal edge image in time series, and finds a time series combination of horizontal edge images that matches the moving direction and moving amount of the reflected edge image estimated in step 4. Then, it is extracted as a time-series combination of reflection edge image candidates (hereinafter also referred to as “reflection edge image candidates”) (S6). Since the edge image mainly moves upward, there is a possibility that an edge corresponding to the lower eyelid may be detected due to a noise factor or the like.

  Subsequently, the face image processing ECU 14 determines whether or not it blinks in time series, and the combination of the time series of the horizontal edge images determined to be blinking is the edge image candidate attributed to the upper eyelid. Extracted as a combination of series (S7). For example, it is determined that a time-series combination of horizontal edge image candidates that move vertically within a predetermined time is blinking in time series. The “predetermined time” is an arbitrary time and is a time that can be determined to be blinking by observing the movement of the horizontal edge image within that time.

  Next, the face image processing ECU 14 adds the time series combination of the reflected edge image candidates extracted in step 6 to the lower eyelid paired with the edge candidate caused by the extracted upper eyelid in step 7. It is determined whether or not there is a resulting edge image candidate, and the edge image candidates resulting from these upper and lower eyelids are excluded from the reflected edge image candidates (S8). For example, an edge image candidate that is linked to the motion of the edge image candidate caused by the left and right upper eyelids and that maintains parallelism is extracted as an edge image candidate caused by the lower eyelid, and is a reflected edge image candidate. Exclude from

  Subsequently, the face image processing ECU 14 sets (detects) the horizontal edge image remaining as the reflection edge image candidate as the edge image reflected on the eyeglass lens, and excludes it from the eye candidates (upper eyelid candidate, lower eyelid candidate). (S9). In the eye detection process executed thereafter, for example, the eye candidate having the highest fitness value is detected as an eye from among the eye candidates.

  According to the face image processing apparatus 10 as described above, the movement direction and the movement amount of the reflection edge projected on the eyeglass lens are estimated based on the acquired movement state of the host vehicle, and the estimated reflection edge is estimated. By excluding the horizontal edge that matches the movement direction and movement amount from the eye candidates, the eye detection accuracy is improved.

  According to such a face image processing apparatus 10, information on the movement of the vehicle is acquired, and based on the acquired information, the movement state of the edge image reflected on the eyeglass lens of the driver D is estimated and estimated. Based on the movement state of the edge image, the edge image actually reflected on the eyeglass lens is detected. Since the edge image resulting from the scenery outside the vehicle moves with the moving state of the vehicle, the moving state of the edge image reflected on the eyeglass lens can be estimated based on the moving state of the vehicle. As a result, an edge image whose position changes in the same manner as the estimated movement state can be excluded as an unnecessary edge from the candidates for detecting the eye, and the eye detection accuracy can be improved. Further, it is possible to detect and remove an unnecessary edge image reflected on the eyeglass lens with a simple configuration by using the hardware of the conventional face image processing device without adding a new dedicated device.

  In addition, the face image processing apparatus 10 does not detect an edge image that moves up and down within a predetermined time as a reflected edge image. When the vehicle moves forward, the reflected edge image moves upward on the eyeglass lens in accordance with the moving direction of the vehicle, so the edge image that moves up and down within a predetermined time is reflected on the eyeglass lens. The possibility of being an edge image is extremely low, and conversely, the possibility of being an eye is high. Therefore, the eye detection accuracy can be improved by not setting the edge image that moves up and down within a predetermined time as the edge image reflected on the eyeglass lens.

  Next, a face image processing apparatus according to a second embodiment of the present invention will be described with reference to FIG. FIG. 4 is a flowchart showing a processing procedure executed by the face image processing ECU according to the second embodiment of the present invention. The difference between the face image processing apparatus of the second embodiment and the face image processing apparatus 10 of the first embodiment is that the processing procedure executed by the face image processing ECU is different. Specifically, step 3 and step 4, the process of a new step 10 is executed. The device configuration of the face image processing device according to the second embodiment is the same as that of the face image processing device 10 of the first embodiment.

  Since the processing of steps 1 to 9 is the same as the processing procedure executed by the face image processing ECU 14 of the first embodiment, the newly added processing of step 10 will be described. In step 10, the face image processing ECU detects the spectacle lens based on the positions of the nose and mouth and identifies the type. Here, for example, the type and shape (size, curvature, etc.) of the spectacle lens are detected based on the captured image information. In addition, it is good also as a structure which recognizes the kind and shape of a spectacle lens by referring the database memorize | stored in the memory | storage part.

  Even in such a face image processing apparatus of the second embodiment, the same operations and effects as the face image processing apparatus 10 of the first embodiment can be obtained. In addition, since the type and shape of the spectacle lens can be recognized in the face image processing apparatus of the second embodiment, the movement state of the reflection edge image is estimated by correcting according to the type and shape of the spectacle lens. The accuracy can be improved. For example, the moving direction of the reflection edge image may differ depending on the curvature of the spectacle lens. Further, when the size of the spectacle lens is different, the number of times of detection of a series of edge images corresponding to a change with time is also different. Therefore, it is possible to improve the detection accuracy of the reflected image by correcting the moving state of the reflected edge image in consideration of the type and shape of the spectacle lens, and to further improve the eye detection accuracy. it can.

  Next, a face image processing apparatus according to a third embodiment of the present invention will be described with reference to FIG. FIG. 5 is a flowchart showing a processing procedure executed by the face image processing ECU according to the third embodiment of the present invention. The face image processing device of the third embodiment is different from the face image processing device 10 of the first embodiment in that the processing procedure executed by the face image processing ECU is different. The new steps 11 and 12 are executed. The device configuration of the face image processing device according to the third embodiment is the same as that of the face image processing device 10 of the first embodiment.

  Since the processing of steps 1 to 9 is the same as the processing procedure executed by the face image processing ECU of the first embodiment, the newly added processing of steps 11 and 12 will be described. In step 11, the face image processing ECU detects a horizontal edge image that becomes an edge image candidate caused by the upper eyelid and / or an edge image candidate caused by the lower eyelid (hereinafter referred to as “upper and lower eyelid image candidate”). It is determined whether or not. If the horizontal edge image that is the upper and lower eyelid image candidate has not been detected, the process proceeds to step 12. If the horizontal edge image that is the upper and lower eyelid image candidate has been detected, the process ends.

  In step 12, the face image processing ECU is set to process as an undetected state in which the eyes (upper and lower eyelids) could not be detected in the subsequent eye detection process. Thereby, the malfunction by an undetected state can be avoided in the latter eye detection process.

  Next, a face image processing apparatus according to a fourth embodiment of the present invention will be described with reference to FIG. FIG. 6 is a flowchart showing a processing procedure executed by the face image processing ECU according to the fourth embodiment of the present invention. The difference between the face image processing apparatus according to the fourth embodiment and the face image processing apparatus 10 according to the first embodiment is that the processing procedure executed by the face image processing ECU is different. The new steps 13 and 14 are executed. The device configuration of the face image processing device according to the fourth embodiment is the same as that of the face image processing device 10 of the first embodiment.

  Since the processing of Steps 1 to 9 is the same as the processing procedure executed by the face image processing ECU 14 of the first embodiment, the newly added processing of Steps 13 and 14 will be described. In step 13, the face image processing ECU determines whether or not the reflected edge image is detected at a position overlapping the pupil. Here, for example, the position of the pupil of the driver D is detected based on the face image of the driver, and it is determined whether or not the position of the pupil is detected at a position where the horizontal edge image overlaps. If the reflected edge image is detected at a position overlapping the pupil, the process proceeds to step 14, and if the reflected edge image is not detected at the position overlapping the pupil, the process is terminated.

  In step 14, the face image processing ECU is set to process as an undetected state in which the pupil cannot be detected in the pupil detection process executed in the subsequent stage. As a result, it is possible to avoid problems in the undetected state in the subsequent pupil detection process. If the position of the edge image that reflects the position of the pupil overlaps, there is a possibility that the detection accuracy of both may be reduced, so if both are present at the position where they overlap, the process is terminated as an undetected state. Thus, a decrease in detection accuracy can be avoided.

  As mentioned above, although this invention was concretely demonstrated based on the embodiment, this invention is not limited to the said embodiment. For example, other techniques may be used as the technique for detecting facial feature points, the technique for detecting the face orientation angle, and the technique for detecting eyes. In short, it is only necessary to estimate the moving state of the reflected edge image based on the moving state of the vehicle and to detect the edge image reflected on the eyeglass lens based on the estimated moving state.

  In the above-described embodiment, the “edge image moving in the vertical direction within a predetermined time” is not detected as a reflection edge image, but the “edge image moving downward” is not detected as a reflection edge image. It is good. Moreover, in the said embodiment, the process of step 7 and 8 does not need to be performed.

  Moreover, in the said embodiment, although the face image processing apparatus 10 is set as the structure provided with the vehicle speed sensor 18, the blinker sensor 20, and the steering sensor 22 as a movement information acquisition means of this invention, it is set as the information from another sensor and a computer. Based on this, the moving state of the host vehicle may be acquired. Further, for example, the movement information may be acquired only by the vehicle speed sensor 18.

  In the third and fourth embodiments, step 10 may be executed, or steps 11 to 14 may be executed.

It is a block diagram which shows the face image processing apparatus which concerns on embodiment of this invention. It is the schematic which shows the time change of the image around an eye, and shows the position change of the reflection edge image reflected on the spectacles lens. It is a flowchart which shows the process sequence performed by face image process ECU which concerns on 1st Embodiment of this invention. It is a flowchart which shows the process sequence performed by face image process ECU which concerns on 2nd Embodiment of this invention. It is a flowchart which shows the process sequence performed by face image process ECU which concerns on 3rd Embodiment of this invention. It is a flowchart which shows the process sequence performed by face image process ECU which concerns on 4th Embodiment of this invention.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Face image processing apparatus, 12 ... Face image pick-up camera, 14 ... Face image processing ECU, 16 ... Column cover, 18 ... Vehicle speed sensor, 20 ... Winker sensor, 22 ... Steering sensor, 24 ... Face feature point extraction part, 26 ... face orientation angle detection unit, 28 ... edge movement direction estimation unit, 30 ... reflection edge detection unit, 32 ... eye detection unit, 40 ... spectacle lens.

Claims (2)

In an image processing apparatus including an image processing unit that acquires a face image of a driver of a vehicle, performs image processing based on the acquired face image, and detects the eyes of the driver.
Movement information acquisition means for acquiring information relating to movement of the vehicle;
Estimating means for estimating a moving state of an edge image reflected on a spectacle lens attached to the driver based on information on movement of the vehicle acquired by the movement information acquiring means;
Reflection detection means for detecting a reflection edge image that is an edge image reflected on the eyeglass lens based on the movement state of the edge image estimated by the estimation means;
Equipped with a,
The image processing means excludes the reflection edge image detected by the reflection detection means from eye candidates,
The face image processing device, wherein the reflection detection means does not detect an edge image moving downward as the reflection edge image . In an image processing apparatus including an image processing unit that acquires a face image of a driver of a vehicle, performs image processing based on the acquired face image, and detects the eyes of the driver.
Movement information acquisition means for acquiring information relating to movement of the vehicle;
Estimating means for estimating a moving state of an edge image reflected on a spectacle lens attached to the driver based on information on movement of the vehicle acquired by the movement information acquiring means;
Reflection detection means for detecting a reflection edge image that is an edge image reflected on the eyeglass lens based on the movement state of the edge image estimated by the estimation means;
Equipped with a,
The image processing means excludes the reflection edge image detected by the reflection detection means from eye candidates,
The face image processing device, wherein the reflection detection means does not detect an edge image that moves in a vertical direction within a predetermined time as the reflection edge image .

Images