JP6455761B2 - Pulse wave detection device and pulse wave detection program - Google Patents

Description

  The present invention relates to a pulse wave detection device and a pulse wave detection program, for example, to a device that detects a pulse wave using image processing.

Detection of a pulse wave is extremely important for grasping the physiological state of human beings, and a technique for detecting this from a remote location without contact is required.
As such a technique, there is a research conducted at Massachusetts Institute of Technology, a non-patent literature.
In this technology, a subject's face is photographed with a web camera, and the pulse wave is detected by analyzing the movie after photographing with a laptop computer (that is, not real-time processing but later processing). .
In this research, pulse waves were detected using special expensive devices such as lasers and Doppler radars, but non-contact detection of pulse waves was realized using inexpensive general-purpose equipment and a simple method. It is a thing.

Here, the outline of this technique will be described with reference to FIG.
First, as shown in FIG. 16A, a rectangular evaluation area 101 is set on the video screen, and the subject's face is seated in the evaluation area 101. Shoot with video. The experiment is performed indoors and sunlight entering through the window is used as the light source.
When the obtained moving image is separated into R component, G component, and B component and averaged, a fluctuation with a pulse wave is obtained as shown in FIG.
Each of these components includes a pulse wave signal that is weighted according to the light absorption characteristics of hemoglobin, and the pulse wave is generated when ICA (Independent Component Analysis) is performed on the pulse wave signal. can get.

  A pulse wave is obtained from a moving image in this way because the volume of the blood vessel changes with the heartbeat of the subject, which changes the optical distance through which sunlight passes through the skin, which changes the reflected light from the face. Because it appears as

  However, the moving image has a variation in pixel characteristics due to camera characteristics that cannot be seen with the naked eye, and this has a problem that it is difficult to detect a pulse due to a disturbance factor.

Advances in Noncontact, Multiparameter Physiological Measurements Using a Webcam, Ming-Zher Poh, Daniel J. McDuff, and Rosalind W.M. Picard, IEEE Transactions on Biomedical Engineering, Vol. 58, no. 1, January 2011

  An object of the present invention is to correct pixel variations caused by camera characteristics in detecting pulse waves.

(1) In order to achieve the above object, according to the first aspect of the present invention, in the first aspect of the present invention, a moving image acquisition unit that acquires a moving image obtained by capturing an area including at least the skin of the subject with a predetermined camera, and the acquisition Skin part specifying means for specifying the skin part of the subject in the moving image, and a predetermined color space component generated in the moving picture due to the characteristics of the camera with respect to pixels of the specified skin part A variation correction unit that corrects the variation of the color space component, a pulse wave acquisition unit that acquires the pulse wave of the subject from the temporal change of the color space component in the corrected skin portion, and outputs the acquired pulse wave And a pulse wave detector characterized by comprising an output means.
(2) In the invention described in claim 2, the moving image acquisition unit acquires a moving image obtained by photographing an area including the target person's face, and the variation correction unit sets a correction target color of the target person's face. The pulse wave detection device according to claim 1, wherein variations in color space components are corrected by limiting to colors.
(3) In the invention according to claim 3, a moving image acquisition means for acquiring a moving image obtained by photographing a region including at least the skin of the subject with a predetermined camera, and the skin of the subject shown in the acquired moving image. Skin portion specifying means for specifying a portion, variation correcting means for correcting a variation in a predetermined color space component generated in the moving image due to the characteristics of the camera, and the color space component in the corrected skin portion A pulse wave acquisition means for acquiring the pulse wave of the subject from the time change of the output, an output means for outputting the acquired pulse wave, and the color space component generated in the skin portion as the skin portion moves And a updating means for updating a correction value used for the correction by performing a predetermined statistical process on the change of the pulse wave.
(4) In the invention according to claim 4, a moving image acquisition means for acquiring a moving image obtained by photographing a region including at least the skin of the subject with a predetermined camera, and the skin of the subject shown in the acquired moving image. Skin portion specifying means for specifying a portion, variation correcting means for correcting a variation in a predetermined color space component generated in the moving image due to the characteristics of the camera, and the color space component in the corrected skin portion A pulse wave acquisition means for acquiring the pulse wave of the subject from the time change of the output, and an output means for outputting the acquired pulse wave, the color space component to be corrected by the variation correction means, and the The color space component used by the pulse wave acquisition unit to acquire the pulse wave is the chromaticity component (Q) of the YIQ color space including the lightness component (Y) and the chromaticity components (I, Q). It is characterized by Providing pulse-wave detecting device.
(5) In the invention according to claim 5, a correction value used for the correction by applying a predetermined statistical process to the change of the color space component generated in the skin part as the skin part moves. The pulse wave detection device according to claim 1, 2, and 4 is provided.
(6) In the invention according to claim 6, the update unit updates the correction value for each pixel constituting the moving image, and the variation correction unit sets a correction value corresponding to the pixel for each pixel. The pulse wave detection device according to claim 3 or 5, wherein the color space component is used to correct the color space component.
(7) In the invention according to claim 7, the update means has a variation in the color space component due to the characteristics of the camera, and at least the change in the color space component due to the pulse wave of the subject. The pulse wave detection device according to claim 3, wherein the update is completed when the value converges to a value smaller than the magnitude of.
(8) In the invention according to claim 8, the variation correction unit corrects the variation using the latest correction value while the update unit updates the correction value, and the pulse wave acquisition unit. The pulse wave detection apparatus according to claim 3, wherein the pulse wave is acquired from the skin portion corrected with the latest correction value. I will provide a.
(9) In the invention described in claim 9, brightness change acquisition means for acquiring a change in brightness caused by a change in the shooting environment of the movie, and correcting the brightness of the movie using the acquired change in brightness. Brightness correction means, wherein the pulse wave acquisition means acquires the pulse wave of the subject from the temporal change of the color space component in the skin portion further corrected by the brightness correction means. A pulse wave detection device according to any one of claims 1 to 8 is provided.
(10) In the invention according to claim 10, the region including at least the skin of the subject includes the face of the subject, any one of claims 1 to 9. A pulse wave detection device according to the claims is provided.
(11) In the invention according to claim 11, the subject is a passenger of a transport device, and comprises a monitoring means for monitoring the physical condition of the passenger using the output pulse wave. A pulse wave detection device according to any one of claims 1 to 10 is provided.
(12) In the invention described in claim 12, a moving image acquisition function for acquiring a moving image obtained by photographing a region including at least the skin of the subject with a predetermined camera, and the skin of the subject shown in the acquired moving image. A skin part specifying function for specifying a part, a dispersion correcting function for correcting a variation in a predetermined color space component generated in the moving image due to the characteristics of the camera with respect to pixels of the specified skin part, and the correction A pulse wave detection function for realizing a pulse wave acquisition function for acquiring a pulse wave of the subject from a time change of the color space component in the skin portion and an output function for outputting the acquired pulse wave by a computer Provide a program.
(13) In the invention described in claim 13, a moving image acquisition function for acquiring a moving image obtained by photographing a region including at least the skin of the subject with a predetermined camera, and the skin of the subject shown in the acquired moving image. A skin part specifying function for specifying a part, a dispersion correcting function for correcting a dispersion of a predetermined color space component generated in the moving image due to the characteristics of the camera, and the color space component in the corrected skin part A pulse wave acquisition function for acquiring the subject's pulse wave from a time change of the object, an output function for outputting the acquired pulse wave, and the color space component generated in the skin part as the skin part moves An update function for updating a correction value used for the correction by performing predetermined statistical processing on the change of the above is provided.
(14) In the invention according to claim 14, a moving image acquisition function for acquiring a moving image obtained by photographing a region including at least the skin of the subject with a predetermined camera, and the skin of the subject shown in the acquired moving image A skin part specifying function for specifying a part, a dispersion correcting function for correcting a dispersion of a predetermined color space component generated in the moving image due to the characteristics of the camera, and the color space component in the corrected skin part A pulse wave detection program for realizing, by a computer, a pulse wave acquisition function for acquiring the pulse wave of the subject from a change in time and an output function for outputting the acquired pulse wave, and correction by the variation correction function color space components used to subject to color space component and the pulse wave acquiring function of acquiring the pulse wave, YIQ color consisting of lightness component (Y), and chrominance components (I, Q) Wherein a chromaticity component (Q) between to provide a pulse wave detection program, characterized in that.

(1) According to the first aspect of the present invention, it is possible to detect a pulse wave satisfactorily by correcting a variation in a predetermined color space component generated in a moving image due to the characteristics of the camera.
(2) According to the second aspect of the present invention, it is possible to detect a pulse wave from a face that is exposed daily and that is easy to shoot.
(3) According to the invention described in claim 3, the correction value is updated by the change of the color space component caused by the movement of the face. At this time, the variation of the correction value can be equalized by statistical processing.
(4) According to the invention described in claim 4, it is possible to improve the accuracy of the correction by correcting the area for each pixel instead of correcting the area with the representative value.
(5) According to the invention described in claim 5, the pulse wave can be detected satisfactorily by converging the correction until the variation of the color space component becomes smaller than the variation of the pulse wave.
(6) According to the invention described in claim 6, it is possible to detect the pulse wave while correcting.
(7) According to the seventh aspect of the present invention, by further correcting the brightness of the moving image using the change in brightness, the robustness against the environmental change is improved, and the pulse wave is generated even in the environment where the brightness changes. Can be detected.
(8) According to the invention described in claim 8, the color space component for pulse wave detection is a Q component that has been found to be suitable for pulse wave detection, thereby providing robustness to disturbance elements. Can be improved.
(9) According to the invention described in claim 9, the physical condition of the passenger of the transport device can be monitored.
(10) According to the invention described in claim 10, the pulse wave detection device can be configured easily and inexpensively by distributing the pulse wave detection program and installing it in a general-purpose computer.

It is the figure which showed the structure of the pulse-wave detection apparatus. It is a figure for demonstrating a color space. It is a figure for demonstrating the mechanism which detects a pulse wave from a moving image. It is a flowchart for demonstrating the procedure of the whole process. It is a flowchart for demonstrating the procedure of a skin color data collection process. It is a flowchart for demonstrating the procedure of a pulse wave detection process. It is a figure showing the result by experiment. It is a figure for demonstrating the correction method of the change of the brightness. It is a flowchart for demonstrating the procedure of the whole process in 2nd Embodiment. It is a flowchart for demonstrating the procedure of an eye color data collection process. It is a flowchart for demonstrating the procedure of a brightness change countermeasure process. It is a figure for demonstrating the dispersion | variation in chrominance. It is a flowchart for demonstrating the procedure of the whole process in 3rd Embodiment. It is a flowchart for demonstrating the procedure of a camera characteristic data update process. It is a flowchart for demonstrating the procedure of a camera characteristic countermeasure process. It is a figure for demonstrating a prior art.

(1) Outline of Embodiment In the first embodiment, the pulse wave detection device 1 color-converts a frame image of a moving image from an RGB component to an HSV component, and the user's skin color prepared in advance as an H component To identify the skin area. The reason why the H component is used is that the robustness is improved when the H component is used for specifying the skin.

Next, the pulse wave detection device 1 converts the skin portion of the frame image into a YIQ component, and uses Qm obtained by averaging the Q values of the pixels as a pulse wave signal. The reason why the Q component is used is that the robustness is improved when the Q component is used for detecting the pulse wave signal.
The pulse wave detection device 1 performs the above processing on each frame image to acquire a time-series change in the pulse wave signal Qm, and outputs this as a pulse wave.

  Thus, since the pulse wave detection device 1 can set the skin part in the moving image as an evaluation region (ROI; Region of Interest), it is preferable to exclude disturbance elements such as the background from the pulse wave detection target. A pulse wave can be detected.

In the second embodiment, the pulse wave detection device 1 color-converts a moving image frame image from an RGB component to a YIQ component, and uses the eye color of the user prepared in advance as the Q component to Is identified.
Then, the pulse wave detection device 1 detects the brightness of the shooting environment based on the Y value of the eye portion. Since no pulse wave signal appears in the eye portion, it can be used as a lightness detection target.

Next, the pulse wave detection device 1 detects the pulse wave signal Qm from the average of the Q values of the skin portions of the frame image, and further subtracts the average value Ye of the Y values of the eye portions from this to determine the brightness. The change is corrected and Qm after brightness correction is output.
Thereby, even when the brightness changes as the user moves with a vehicle or the like, the pulse wave can be detected well.

In the third embodiment, the pulse wave detection device 1 includes camera characteristic data for correcting variations in pixel characteristics due to camera characteristics. Then, the Q value of the skin portion of the frame image is corrected by the camera characteristic data.
The pulse wave detection device 1 updates the camera characteristic data by using the change in the Q value of the skin portion that is generated when the skin portion moves in the screen by the user's movement.
Further, the pulse wave detection device 1 limits the color to be corrected to the skin color, which eliminates the need for complicated algorithms and calculations, thereby reducing the calculation load and performing the real-time processing of the video well. be able to.

(2) Details of the embodiment (first embodiment)
FIG. 1 is a diagram showing a configuration of a pulse wave detection device 1 according to the present embodiment.
The pulse wave detection device 1 is mounted on a vehicle, for example, and monitors a pulse wave of a passenger (such as a driver or a passenger in a passenger seat) and grasps a physiological state such as a physical condition and a tension state of the driver.
It can also be used to detect and monitor pulse waves of patients and victims at medical sites and disaster sites.

  The pulse wave detection device 1 includes a CPU (Central Processing Unit) 2, a ROM (Read Only Memory) 3, a RAM (Random Access Memory) 4, a display unit 5, an input unit 6, an output unit 7, a camera 8, a storage unit 9, and the like. It detects (or estimates) the pulse wave of the user 10 (the subject of pulse wave detection).

The CPU 2 is a central processing unit that performs various types of information processing and control in accordance with programs stored in the storage unit 9 and the ROM 3.
In the present embodiment, the moving image captured by the camera 8 is subjected to image processing to detect the pulse wave of the user 10.

The ROM 3 is a read-only memory, and stores basic programs and parameters for operating the pulse wave detection device 1.
The RAM 4 is a readable / writable memory, and provides a working memory when the CPU 2 operates.
In the present embodiment, by developing and storing a frame image (one frame of still image) constituting a moving image or storing a calculation result, the CPU 2 causes the skin portion of the frame image (hereinafter referred to as skin). Assist in detecting pulse wave from part).

The display unit 5 is configured using a display device such as a liquid crystal screen, and displays information necessary for operation of the pulse wave detection device 1 such as an operation screen of the pulse wave detection device 1 and display of a pulse wave.
The input unit 6 is configured using an input device such as a touch panel installed so as to be superimposed on the display device, and receives input of various types of information based on whether or not the screen display is touched.

The output unit 7 is an interface that outputs various types of information to an external device. For example, the output unit 7 outputs a detected pulse wave, outputs a pulse obtained from the pulse wave, or changes in the pulse wave appear. An alarm can be output in case.
Moreover, the output part 7 can output to other control apparatuses, such as a control apparatus which controls a vehicle. The control device that has received the pulse wave or pulse output from the output unit 7 determines, for example, the driver's drowsiness or tension (described later) and controls the driver, for example, Controls that vibrate the seat, output of warning sounds and messages, etc. can be performed. In addition, as control for the vehicle, at least one of inter-vehicle distance control, vehicle speed control, or brake control can be performed in accordance with the tension state of the driver determined based on the pulse wave. For example, when the control device determines that the driver is in a high tension state exceeding a predetermined value, the control device controls the inter-vehicle distance to be larger than the reference value, and controls the vehicle speed to be equal to or lower than the predetermined vehicle speed. If the vehicle speed is higher than the predetermined vehicle speed, a deceleration process by an automatic brake operation is performed.

  The camera 8 is configured using an optical system composed of lenses and an image sensor that converts an image formed thereby into an electrical signal, and is installed so that the vicinity of the face of the user 10 becomes a shooting screen. Yes.

Although an expensive camera can be used as the camera 8, the pulse wave detection device 1 uses a general-purpose product such as a web camera.
In the pulse wave detection device 1, the pulse wave can be well detected even by the general-purpose camera 8, so that the cost can be reduced.

The camera 8 captures the subject at a predetermined frame rate and outputs a moving image composed of these continuous frame images (still images).
The frame image is composed of an array of pixels (pixels), which is the minimum unit constituting the image, and each pixel is arranged by color components (R value, G value, B value) in the RGB space.

  The storage unit 9 is configured using a storage medium such as a hard disk or an EEPROM (Electrically Erasable Programmable Read-Only Memory), and stores a program and data for the CPU 2 to detect a pulse wave.

The storage unit 9 stores a pulse wave detection program 12, a user database 14, camera characteristic data 15, and the like.
Of these, the camera characteristic data 15 is data used in the third embodiment, and will be described later.

The pulse wave detection program 12 is a program that causes the CPU 2 to perform a pulse wave detection process.
The CPU 2 executes the pulse wave detection program to identify the user's skin part in the moving image and detect the pulse wave from the identified skin part.

The user database 14 is a database in which users who use the pulse wave detection device 1 are registered.
In the user database 14, registration data is stored for each user, such as user 1, user 2,.
In the registration data, user-specific information such as face data, skin color data, eye color data,... Is registered.

The face data is obtained by converting the features of the user's face into data, and is used for recognizing and identifying a user seated in front of the camera 8.
The skin color data is data serving as a skin color reference for specifying the user's skin portion in the frame image. The skin portion is specified by comparing the frame image with the skin color data.
The eye color data is data used in the second embodiment and will be described later.

FIG. 2 is a diagram for explaining a color space.
Since the pulse wave detection device 1 converts the color space of the frame image (referred to as color conversion) when detecting the pulse wave, this will be described first.
In general, a color image is represented by three color components. More specifically, the color information associated with each pixel of the image is expressed as a coordinate value of a point in a color space spanned around three color components.
A general-purpose video camera often uses R, G, and B components in the RGB space as color components. Even in the prior art, pulse waves are detected using the R, G, and B components included in the video signal as they are. Is going.

In contrast, the inventor of the present application repeated trial and error and searched for a color component that is more robust (resistant) to disturbance elements.
As a result, it was found that the H component in the HSV space is suitable for specifying the skin portion, and the Q component in the YIQ space is suitable for detecting the pulse wave.
Therefore, in the pulse wave detection device 1, color components are properly used according to the purpose.
Thus, since the reflection characteristics of light differ depending on the observation object, robustness against disturbance can be enhanced by selecting an optimal combination according to the object.

FIG. 2A shows the RGB space.
The RGB space is composed of R, G, and B axes that are orthogonal to each other and represent RGB components.
In the RGB space, color information is represented by an R value (red), a G value (green), and a B value (blue), and the RGB value of a pixel is defined by the coordinate value of a point in the RGB space.
The RGB method is the most common color model, and the camera 8 also outputs a moving image by the RGB method.

FIG. 2B is a diagram showing the HSV space.
The HSV space is represented by a cone having a circular bottom surface, the rotation angle in the direction of the cone surface represents the H component, the distance from the center at the bottom surface represents the S component, and the distance of the perpendicular drawn from the apex of the cone to the bottom surface is V component is represented.
In the HSV space, color information is represented by an H value (color), an S value (saturation), and a V value (lightness), and the HSV value of a pixel is defined by the coordinate value of a point in the HSV space.
The HSV method is mainly used in computer graphics.

  The HSV space can be represented by a cylinder in addition to the cone shown in FIG. In this case as well, the hue (H) changes along the outer periphery of the cylinder, the saturation (S) changes with the distance from the center, and the lightness (V) changes from the apex to the bottom, as in the case represented by the cone. Change towards.

FIG. 2C shows the YIQ space.
The YIQ space includes a Y axis, an I axis, and a Q axis that represent YIQ components and are orthogonal to each other.
In the YIQ space, color information is represented by a Y value (lightness), an I value (chromaticity: warm color system), and a Q value (chromaticity: cold color system), and the YIQ value of a pixel is defined by the coordinate value of a point in the YIQ space. Is done.
The Y value takes a positive value, and the I value and the Q value can take positive and negative values.
The YIQ method is mainly used in video equipment as a component signal method for generating an NTSC signal.

FIG. 3 is a diagram for explaining a mechanism for detecting a pulse wave from a moving image.
FIG. 3A is a diagram illustrating a method in which the pulse wave detection device 1 collects skin color data from the user's face.
The pulse wave detection device 1 captures a still image 30 of the user's face with the camera 8, detects the nose, and sets the nose region 20. The detection of the nose is performed using a general face recognition technique.
Then, the pulse wave detection device 1 converts the color space of the nose region 20 from the RGB space to the HSV space, and generates skin color data from the H value of each pixel.

The reason why the skin color data is collected is the nose region 20 because it is easy to identify by face recognition and the standard skin color is exposed.
In addition to the nose, for example, skin color data can be collected from other areas such as the forehead and cheeks.

FIG. 3B is a diagram for explaining a method for extracting a pulse wave signal from the frame image 31.
In the frame image 31, in addition to the skin portion, hair 21, eyebrows 22, eyes 23, lips 24, background 25, and the like are imprinted.
These parts other than the skin do not include the pulse wave signal or are not suitable for detecting the pulse wave signal, and act as disturbance elements that cause a decrease in accuracy in the pulse wave detection process.

Therefore, the pulse wave detection device 1 performs color conversion on the frame image 31 to generate an HSV image 32, and specifies a portion corresponding to the skin color data as the skin portion 26.
The skin portion 26 is specified on a pixel basis, and all portions where the skin such as the neck is exposed are specified.
As described above, the pulse wave detection accuracy can be improved by removing the disturbance element and securing the maximum portion including the pulse wave signal.

The pulse wave detection device 1 extracts the skin part in the frame image 31 from the position of the skin part 26 in the HSV image 32 and converts it into a YIQ image. As a result, a skin portion 26a in the YIQ space is obtained.
The pulse wave detection device 1 calculates Qm by averaging the Q values of the pixels of the skin portion 26a, and outputs Qm as a pulse wave signal.

  In the present embodiment, the skin image 26a is obtained by converting the frame image 31 into a YIQ image. However, the skin image 26a can also be obtained by converting the skin portion 26 of the HSV image 32 into a YIQ image. .

FIG. 3C is a diagram showing a pulse wave detected from the moving image.
The pulse wave detection device 1 forms the pulse wave 35 by arranging the pulse wave signals Qm output from the respective frame images in time series (in the order of the frame images).
As described above, the pulse wave detection device 1 can detect the pulse wave while tracking and extracting the skin part in accordance with the movement of the user by specifying the skin part of the user in each frame image. Thereby, the following features are obtained.

(Compatibility with real-time processing)
In the prior art, a moving image is shot while the subject is stationary with the face in the evaluation area 101. Since there is a possibility that the face may deviate from the evaluation area 101 during shooting, the moving image that has been shot is analyzed.
On the other hand, since the pulse wave detection device 1 specifies the skin portion 26 in each frame image of the moving image, the evaluation area moves within the moving image while being fixed to the user's skin.
Even if the face moves, the skin portion does not deviate from the evaluation area, so that the pulse wave can be detected in real time.

Further, if the user's face is recognized and tracked with each frame image, the computer needs a high processing capability and the possibility that the face recognition will fail is high.
On the other hand, the pulse wave detection device 1 can track the skin portion by a simple process of identifying the skin portion by comparing the HSV image with the skin color data. Therefore, it is suitable for real-time processing.

Further, since the pulse wave is detected by simple and low-load processing, frame image dropout (so-called frame dropout) caused by the processing capability of the computer can be suppressed even when real-time processing is performed.
Since the interval between the processed frame images is the pulse wave measurement interval (sampling rate), it is possible to prevent the measurement interval from being widened by preventing frame dropping.
As a result, a high temporal resolution of the pulse wave can be maintained, and the detection accuracy of the pulse wave is improved.

Also, face recognition processing is only performed when skin color data is registered, and since skin color data is already registered, face recognition fails at the site, skin color data cannot be collected, and pulse detection cannot be performed. Can be avoided, and the reliability of the measurement is improved.
The detection of the pulse wave is mainly used when the current physiological state of the subject is monitored, and it is important that it can be processed in real time.

(Background removal)
In the prior art, since it is difficult to match the evaluation area 101 with the shape of the face of the subject, a disturbance element such as a background enters the evaluation area 101, and the pulse wave detection accuracy may be reduced.
On the other hand, in the pulse wave detection device 1, since the evaluation region and the skin portion 26 always coincide with each other, it is possible to prevent disturbance elements such as a background other than the face from entering the evaluation region. Detection can be performed.

(Removal of facial parts not related to pulse wave)
In the prior art, even when the subject's face is correctly set in the evaluation area 101, the face portions (hair, eyes, mouth, etc.) that are not related to the pulse wave are in the evaluation area. It may become a disturbance factor and affect the detection accuracy of the pulse wave.
On the other hand, in the pulse wave detection device 1, since these facial elements are excluded from the skin portion 26, the pulse wave detection accuracy can be increased.
Furthermore, even if the user blinks or opens / closes the mouth, the skin portion 26 is dynamically set in accordance with the movement of the face, so that disturbance elements due to the movement of the face can be excluded from the evaluation region. .

FIG. 4 is a flowchart for explaining an overall processing procedure performed by the pulse wave detection device 1.
The following processing is performed by the CPU 2 in accordance with the pulse wave detection program 12.
First, when the CPU 2 detects that the user is seated at the specified position from the change in the image of the camera 8, the CPU 2 acquires an image of the user's face and stores it in the RAM 4 (step 5).
This image may be taken as a still image or taken out from a frame image of a moving image.
As described above, the pulse wave detection device 1 includes moving image acquisition means for acquiring a moving image obtained by photographing at least a region including the skin of the subject, and the region including the skin includes the face of the subject.

Next, the CPU 2 recognizes the face image stored in the RAM 4 and searches the face image by comparing it with the face data in the user database 14 (step 10).
When the face is searched, the CPU 2 determines that the user has been registered (step 15; Y), acquires the user's skin color data from the user database 14, and stores it in the RAM 4 (step 20).

On the other hand, when the face is not searched, the CPU 2 determines that the user is not registered (step 15; N), performs skin color data collection processing, and collects skin color data from the face image ( Step 25).
Then, the pulse wave detection device 1 forms face data from the face image, and stores the face data and the skin color data in association with each other in the user database 14 to perform user registration (step 27).

When acquiring the skin color data, the CPU 2 acquires a frame image from the moving image transmitted from the camera 8 and stores it in the RAM 4 (step 30).
Next, the CPU 2 performs a pulse wave detection process for detecting a pulse wave from the frame image stored in the RAM 4 (step 35).

Next, the CPU 2 determines whether or not to continue detecting the pulse wave (step 40).
When the detection of the pulse wave is continued (step 40; Y), the pulse wave detection device 1 returns to step 30 and performs the pulse wave detection process on the next frame image of the moving image.
On the other hand, when the user does not continue detecting the pulse wave by pressing the stop button or the like (step 40; N), the pulse wave detection device 1 ends the process.

FIG. 5 is a flowchart for explaining the procedure of the skin color data collection process in step 25.
First, the CPU 2 reads a face image from the RAM 4, recognizes the face (step 50), and detects the nose (step 55).
Next, the CPU 2 sets a nose region for collecting skin color data in the detected nose (step 60).

  As described above, the pulse wave detection device 1 specifies a predetermined region (nasal region) where the skin of the face is exposed by the face image acquisition unit that acquires an image obtained by photographing the face of the subject and the face recognition process. An area specifying means is provided.

Next, the CPU 2 performs color conversion of the color space of the nose region from the RGB space to the HSV space (step 65), and acquires the H value of each pixel (step 70).
Next, the CPU 2 calculates the Hm by averaging the H value of each pixel (step 75), and further calculates the standard deviation σ of the H value (step 80).

Next, the CPU 2 calculates a lower limit value Hlo = Hm−n × σ of the H value from Hm and σ and stores it in the RAM 4 (step 85). n will be described later.
Further, the CPU 2 calculates the upper limit value Hhi = Hm + n × σ of the H value and stores it in the RAM 4 (step 90).
Then, the CPU 2 outputs the lower limit value and upper limit value of the H value as skin color data (step 95), and returns to the main routine (FIG. 5).

The output skin color data (Hlo and Hhi) function as a reference component that is a color space component that serves as a reference for specifying the skin portion of the subject, and is registered by user registration in step 27 (FIG. 4). The
As described above, the pulse wave detection device 1 includes reference component registration means for registering a reference component, and registers the color space component of the nose region by performing statistical processing using the average value and the standard deviation.

n is a multiplier of σ and defines the width of the H value centered on Hm. As will be described later, since the pulse wave detection device 1 specifies a portion where the H value is between these widths as a skin portion from the frame image, n can be adjusted to an appropriate value through experiments or the like.
For example, when n = 3, a portion having an H value in the range of Hm ± 3σ is specified as a skin portion.

FIG. 6 is a flowchart for explaining the procedure of the pulse wave detection process in step 35 (FIG. 5).
First, the CPU 2 converts the color space of the frame image stored in the RAM 4 from the RGB space to the HSV space, and stores the converted HSV image in the RAM 4 (step 100).
Next, the CPU 2 sets a counter i for counting the order of pixels to i = 0 (step 103).

Next, the CPU 2 acquires Hi, which is the H value of the i-th pixel among the pixels of the HSV image stored in the RAM 4 (step 105).
Then, the CPU 2 determines whether Hi satisfies Hlo <Hi <Hhi, that is, whether Hi is within the range of the skin color data (step 110).
When the Hi is within this range, the CPU 2 determines that the pixel corresponds to the skin color data.
When Hi satisfies the inequality, that is, when Hi corresponds to the skin color data (step 110; Y), the CPU 2 stores the position coordinates of the pixel in the RAM 4 (step 115).

  After storing the position coordinates in step 115, or if Hi is not in the range of the skin color data in step 110 (step 110; N), has the CPU 2 made the determination in step 110 for all the pixels of the HSV image? It is determined whether or not (step 120).

If there is a pixel that has not yet been determined (step 120; N), the CPU 2 increments i by 1 and updates it to i = i + 1 (step 123), returns to step 105, and the same for the next pixel. Repeat the process.
By performing the above steps 100 to 123, the position coordinates corresponding to the skin portion are stored in the RAM 4.

Next, the CPU 2 acquires the evaluation target pixel in the frame image by specifying the pixel located at the position coordinate stored in the RAM 4 in the frame image (step 125).
As described above, the pulse wave detection device 1 includes a skin part specifying unit that specifies a skin part (evaluation target pixel) of the subject shown in the moving image.
This specification is performed by setting a portion corresponding to a reference component registered in a predetermined color space component in a moving image as a skin portion in units of pixels.

Next, the CPU 2 converts the color space of the evaluation target pixel from the RGB space to the YIQ space (step 130).
Then, the CPU 2 averages the Q values of the respective pixels to calculate an average value Qm (step 135), outputs this as a pulse wave signal (step 140), and returns to the main routine (FIG. 4).

A pulse wave signal is carried on the Q value of the pixel to be evaluated. By averaging the pulse wave signal to obtain Qm, the influence of noise can be reduced.
Although the pulse wave signal Qm is detected from one frame image by the above processing, if this is performed for each successive frame image and the pulse wave signal Qm is arranged in the order of the frame image, the pulse wave signal Qm is changed depending on the time change. The pulse wave shown in 3 (c) is obtained.
As described above, the pulse wave detection device 1 includes a pulse wave acquisition unit that acquires a pulse wave from a time change of a predetermined color space component in the skin portion, and an output unit that outputs the pulse wave.

As described above, the pulse wave detection device 1 performs identification of the skin portion with the H component in the HSV space and performs pulse wave detection with the Q component in the YIQ space.
For this reason, the color space component used for the skin portion specifying means to specify the skin portion and the color space component used for the pulse wave acquisition means to acquire the pulse wave are different color space components.
The pulse wave detection device 1 includes color space conversion means for converting the color space of the moving image, and the skin part specifying means and the pulse wave acquisition means acquire the color space component in the color space converted by the color space conversion means. doing.
Furthermore, the pulse wave detection device 1 can be provided with monitoring means for monitoring the physical condition of the passenger of the transport device by the pulse wave.

  In this embodiment, after extracting the skin part from the frame image, the extracted skin part is color-converted from the RGB space to the YIQ space. However, the skin part is extracted after color-converting the entire frame image to the YIQ space. You may make it do.

(Second Embodiment)
In the prior art, a pulse wave is detected under a stable brightness by sunlight that is inserted through a laboratory window.
On the other hand, when the pulse wave detection device 1 is used in a vehicle or a medical field, there are various kinds of imaging environments, and it is expected that the brightness changes particularly during the pulse wave detection. In particular, when detecting a pulse wave of a driver or a passenger in a vehicle, a change in brightness is likely to occur frequently due to a change in the traveling position or orientation of the vehicle, a time zone, or the like.
Therefore, it was examined whether or not the detection result is affected by a change in brightness that occurs when the pulse wave detector 1 is actually used. That is, the present inventor tried to change the brightness by making a shadow on the face of the subject with a fan while detecting the pulse wave under illumination with a fluorescent lamp.

FIG. 7 is a diagram showing the results of the experiment.
FIG. 7A shows a time change of the pulse wave signal Qm when the brightness of the environment does not change.
As shown in the figure, the pulse wave is clearly detected.

On the other hand, FIG. 7B shows a time change of the pulse wave signal Qm when only the brightness is changed by making a shadow with a fan on the face of the subject.
The period 41 is a period without a shadow, and the period 42 is a period with a shadow.
As shown in the figure, the pulse wave signal Qm changes greatly with the change in brightness, and the change in Qm is buried in this change, making it difficult to detect the pulse wave.

When the present inventor researched on such a problem, it was found that the pulse wave signal Qm does not appear in the eye part (hereinafter, eye part).
From the skin portion, a pulse wave signal Qm in which a change in lightness is superimposed is detected, and from the eye portion, a change in lightness (or magnitude of lightness) not including a pulse wave is detected. By subtracting the latter, changes in brightness can be corrected.
Further, since the present inventor has found that the Y component of the YIQ space is suitable for detecting the change in brightness, in the present embodiment, the change in brightness is detected by the Y component.
Furthermore, since the pulse wave signal Qm and the Y value of brightness belong to the same color space, they may be simply subtracted.

Each figure of FIG. 8 is a figure for demonstrating the correction method of the change of the brightness.
FIG. 8A is a diagram showing an eye portion 45 used for detecting a change in brightness.
The eye portion 45 is composed of a pupil portion 48 having a deep color and a substantially central portion, an iris portion 47 around the pupil portion 48, and a white eye portion 46 which is close to white and further outside the iris portion 47.

The pulse wave detection device 1 registers the minimum value of the Q value in the eye portion 45 as Qlo, the maximum value of the Q value in the eye portion 45 as Qhi, and these as eye color data.
Note that the pulse wave detection device 1 may register the Q value of the white eye portion 46 as Qlo and the Q value of the pupil portion 48 as Qhi, and register these as eye color data.
As will be described later, the pulse wave detection device 1 uses the eye color data to extract the eye part 45 from the eye area of the user's face, and detects a change in brightness from the change in the Y value of the extracted eye part 45. .

FIG. 8B is a diagram showing the lightness signal 51.
The pulse wave detector 1 averages the Y values detected from the eye portion 45 and generates a lightness signal Ym. When this is plotted in time series, the brightness signal 51 is obtained.
In the example shown in the figure, since a shadow is made on the face in the period 42, the lightness in the period 42 is smaller than that in the period 41.

FIG. 8C is a diagram showing the pulse wave signal 52 before correction.
The pre-correction pulse wave signal 52 is a time-series plot of the pulse wave signal Qm before correcting the change in brightness.
In the pulse wave signal 52 before correction, as shown in FIG. 8C, the pulse wave signal Qm in the period 42 is also lowered due to the influence of the decrease in brightness.

FIG. 8D is a diagram showing the corrected pulse wave signal 53.
The pulse wave detection device 1 generates a corrected pulse wave signal 53 by subtracting the lightness signal Ym from the pulse wave signal Qm before correction. When this is plotted in time series, a corrected pulse wave signal 53 is obtained.
In the post-correction pulse wave signal 53, since the influence due to the change in brightness is removed, an appropriate pulse wave can be obtained even in the period 42 in which the brightness is lowered.

FIGS. 8E and 8F are diagrams for explaining a method in which the pulse wave detection device 1 identifies the eye portion 45 with a moving frame image.
First, as illustrated in FIG. 8E, the pulse wave detection device 1 performs face recognition processing on the frame image and extracts an eye region 55 including the eye portion 45.
Then, the pulse wave detection device 1 extracts the eye portion 45 by applying the eye color data to the eye region 55.

The reason why the eye portion 45 is extracted from the eye region 55 in this way is as follows.
When the eye color data is applied to the frame image, for example, a part corresponding to the eye color data is accidentally extracted in the background or the like, whereas the eye area 55 is composed of a skin part and an eye part 45, This is because only the eye portion 45 corresponds to the color data, and thus the eye portion 45 can be specified reliably.

  Although it is possible to directly specify the eye portion 45 by face recognition, it is necessary to improve the accuracy of face recognition and the failure rate also increases. Thus, in the image recognition, the eye region 55 is roughly set as described above. Decided to identify.

FIG. 9 is a flowchart for explaining an overall processing procedure performed by the pulse wave detection device 1.
The same steps as those in the first embodiment are denoted by the same step numbers, and description thereof is simplified or omitted.
When the user is a registered user (step 15; Y), the CPU 2 acquires skin color data from the user database 14 (step 20), and further acquires eye color data (step 150).

  On the other hand, if the user is not a registered user (step 15; N), the CPU 2 performs skin color data collection processing (step 25), and further performs eye color data collection processing (step 155) to obtain face data and skin color data. The user registration is performed by storing the eye color data in the user database 14 (step 27).

The CPU 2 acquires a frame image (step 30), performs a pulse wave detection process (step 35), and detects a pulse wave signal Qm before correction.
Next, the CPU 2 performs brightness change countermeasure processing on the pulse wave signal Qm before correction (step 160), and outputs the pulse wave signal Qm after correction.

FIG. 10 is a flowchart for explaining the procedure of eye color data collection processing in step 155.
First, the CPU 2 recognizes the face image used in the skin color data collection process (step 180) and detects an eye part (step 185).
As described above, the pulse wave detection device 1 includes an area specifying unit that specifies the area of the eye of the face (in this case, the eye portion) by performing face recognition processing on the face image.
Next, the CPU 2 sets the eye portion as an evaluation area (step 190), and converts the color space of all pixels included in the eye portion from the RGB space to the YIQ space (step 200).

Next, the CPU 2 acquires a Q value for each pixel included in the eye portion and stores it in the RAM 4 (step 205).
Then, the CPU 2 sets the lowest Q value stored in the RAM 4 to Qlo (step 210), and further sets the highest value to Qhi (step 215). The data is registered in the user database 14 in association with the data (step 220), and the process returns to the main routine (FIG. 9).

As described above, the pulse wave detection device 1 includes reference component registration means for registering the color space component of the specified region as reference components (Qlo and Qhi) serving as a reference for specifying the eye portion.
The reference component registration unit registers a value obtained by performing statistical processing for specifying the minimum value and the maximum value on the distribution of the color space component in the specified region.

FIG. 11 is a flowchart for explaining the procedure of the lightness change countermeasure process in step 160.
First, the CPU 2 of the pulse wave detection device 1 performs face recognition using the frame image stored in the RAM 4 to detect a face (step 230), further detects an eye area (step 235), and detects the detected eye area. Is set in the evaluation area (step 240).
Next, the CPU 2 converts the color space of the eye area from the RGB space to the YIQ space and stores it in the RAM 4 (step 245).

Next, the CPU 2 sets a counter j to 0 (step 250), and acquires Qj, which is the Q value of the jth pixel in the eye area, from the RAM 4 (step 253).
Then, the CPU 2 determines the magnitude relationship between Qj, Qlo, and Qhi (step 255).
If Qj is Qlo <Qj <Qhi (step 255; Y), the CPU 2 determines that the pixel is included in the eye portion, acquires the Y value of the pixel, and stores it in the RAM 4 (step 260).

After obtaining the Y value or when Qj does not satisfy Qlo <Qj <Qhi (step 255; N), the CPU 2 determines whether or not all the pixels have been determined (step 265).
If there is a pixel that has not yet been determined (step 265; N), the CPU 2 increments j by one and updates it to j = j + 1 (step 270), and then returns to step 253 for the next pixel. Similar processing is repeated.

By the above processing, the Y value of each pixel of the eye part excluding the skin part from the eye region is obtained.
As described above, the pulse wave detection device 1 specifies the portion of the eye that appears in the moving image in units of pixels by specifying the portion corresponding to the reference component registered in the predetermined color space component in the moving image as the eye portion. The eye part specifying means is provided.

When the CPU 2 stores the Y values of all the pixels in the RAM 4 in this way, the CPU 2 averages them and calculates an average value Ye (step 275).
The average value Ye corresponds to the brightness of the shooting environment, and the difference between this and Ye of the preceding and following frame images represents the change in brightness.
Therefore, the change in brightness can be corrected by subtracting the average value Ye of each brightness from each frame image.
As described above, the pulse wave detection device 1 includes brightness change acquisition means for acquiring a change in brightness caused by a change in the shooting environment of a moving image from a change in a predetermined color space component of the eye portion.

Next, the CPU 2 acquires the pulse wave signal Qm before correction (step 276), subtracts the average value Ye from this to calculate the corrected pulse wave signal Qm (step 277), and calculates the corrected pulse wave signal after correction. The pulse wave signal Qm is output (step 278), and the process returns to the main routine (FIG. 9).
In this way, the pulse wave detection device 1 detects a pulse wave from a lightness correction unit that corrects the lightness of a moving image in units of pixels by using a change in lightness, and a temporal change of a predetermined color space component in the corrected skin portion. Pulse wave acquisition means for acquiring is provided.
Further, the pulse wave detection device 1 acquires a change in brightness with the Y component, specifies a skin portion with the H component, detects a pulse wave with the Q component, and performs processing with different color components.
The pulse wave detection device 1 includes color space conversion means for converting these color spaces.

According to the second embodiment described above, the following effects can be obtained.
(1) A pulse wave can be detected even when the brightness changes due to a change in photographing environment such as a change in external light or a user's movement.
(2) While detecting a pulse wave from the skin of the face, it is possible to detect a change in brightness from the eyes at the same time.
(3) The change in brightness can be corrected without a special device.
In this embodiment, the skin portion is extracted from the frame image and then the brightness of the skin portion is corrected. However, after the brightness correction is performed on the entire frame image, the skin portion is extracted. It may be.

(Third embodiment)
For example, when a general-purpose camera 8 is used, it is not known when a human appreciates a moving image, but there is a variation in characteristics of each pixel that is an obstacle to detecting a pulse wave.
In this embodiment, since the pulse wave is detected by the color component, it is affected by variations in chrominance (color quality) characteristics.

FIG. 12 is a diagram for explaining variation in chrominance due to camera characteristics.
FIG. 12A is a diagram showing the variation in the chrominance characteristics of the pixels of the camera 8 in shades.
Since the chrominance characteristics are not uniform in this way, when the user moves within the screen, the chrominance value changes, which affects the accuracy of pulse wave detection.

FIG. 12B is a diagram comparing pulse waves detected in the left region 61, the center region 62, and the right region 63 of the screen by having the subject move within the screen.
As shown in the figure, where the pulse waves should be detected at the same level, there is a difference in the level due to the difference in chrominance.

Conventionally, correction of the chrominance characteristics of a camera has been performed for various colors by, for example, an expert using a calibration chart. This requires specialized knowledge, takes time, and is difficult for general users.
In addition, since various colors are corrected, the correction process is complicated, and when the pulse wave is processed in real time, frame dropping may occur and the frame rate related to the process may decrease.

Therefore, in the present embodiment, the correction target color is limited to the color of the user's face, and an average value of chrominance for each pixel generated by the user's movement in the screen is stored as camera characteristic data.
An accurate pulse wave can be detected by correcting the change in chrominance using the camera characteristic data.

Further, although the skin portion sweeps (sweeps) the screen as the user moves within the screen, the pulse wave detection device 1 sequentially corrects pixels in the region where the skin portion has been swept.
As described above, in the present embodiment, a correction value can be automatically created in accordance with the movement of the user in the screen.

FIG. 13 is a flowchart for explaining an overall processing procedure performed by the pulse wave detection device 1.
The same steps as those in the first embodiment and the second embodiment are denoted by the same step numbers, and description thereof is simplified or omitted.
After step 5 to step 30, after performing the pulse wave detection process (step 35), the CPU 2 performs a camera characteristic data update process (step 305).

Then, the CPU 2 acquires the latest updated camera characteristic data (step 310), and uses this to correct the pulse wave signal Qm by camera characteristic pair processing (step 315).
Thereafter, the CPU 2 performs brightness change countermeasure processing on the pulse wave signal Qm that has been corrected by the camera characteristics (step 160).

FIG. 14 is a flowchart for explaining the procedure of the camera characteristic data update process in step 305 (FIG. 13).
Steps 500 to 530 are the same as steps 100 to 130 in FIG.
That is, the CPU 2 designates a pixel corresponding to the skin portion in the frame image as an evaluation target pixel, and converts the color space from the RGB space to the YIQ space ( step 530 ).

Next, the CPU 2 selects a pixel at the position coordinate (x, y) among the evaluation target pixels (the selection method may be performed by an arbitrary algorithm), acquires the Q value, and stores it in the RAM 4 ( Step 570 ).
Next, the CPU 2 acquires the latest correction value (that is, the latest value up to the previous correction) Qc (x, y, k) for the position coordinate from the camera characteristic data 15 and stores it in the RAM 4 ( step 575 ). .
Here, k is a counter set for each pixel, and is a parameter representing the number of corrections of the pixel up to the previous time.

Next, the CPU 2 uses these values stored in the RAM 4 to calculate a new Qc value by the following equation, and stores the result in the RAM 4 ( step 580 ).

  Qc (x, y, k + 1) = {Qc (x, y, k) × k + Q (x, y)} / (k + 1)

Next, the CPU 2 updates Qc (x, y, k) of the camera characteristic data 15 with Qc (x, y, k + 1) ( step 585 ).
As described above, since the pulse wave detection device 1 updates the correction value for each pixel using the statistical processing according to the above formula, the pulse wave detection device 1 is predetermined with respect to a change in the color space component that occurs in the skin portion as the face moves. An update means for updating the correction value for each pixel by performing the statistical processing is provided.

Next, the CPU 2 determines whether or not the Qc values have been updated for all the evaluation target pixels ( step 590 ).
If there is a pixel that has not been updated yet ( step 590 ; N), the CPU 2 selects the next pixel to be evaluated (x, y) ( step 595 ), and then returns to step 570 to update all the pixels. If so ( step 590 ; Y), the updating process is terminated and the process returns to the main routine (FIG. 13).

FIG. 15 is a flowchart for explaining the procedure of the camera characteristic countermeasure process in step 315 (FIG. 13).
The CPU 2 calculates Qa (Q value corrected by the camera characteristic countermeasure) by subtracting Qc from the Q value for each pixel set as the evaluation target pixel in the frame image (step 380), and calculates an average value of Qa Then, this is set as the pulse wave signal Qm (step 385), and the process returns to the main routine.

As described above, the pulse wave detection device 1 is corrected with the variation correction unit that corrects the variation of the predetermined color space component generated in the moving image due to the characteristics of the camera using the correction value corresponding to the pixel for each pixel. A pulse wave acquisition means for acquiring a pulse wave from a time change of a color space component in the skin part, and an output means for outputting the pulse wave.
Further, the pulse wave detection device 1 updates the correction value by performing the camera characteristic data update process in step 305 (FIG. 13) and the camera characteristic countermeasure process in step 315 (FIG. 13) in the same loop process. In the meantime, the variation is corrected using the latest correction value, and the pulse wave is acquired from the skin portion corrected with the latest correction value.

As described above, the pulse wave detection device 1 outputs the pulse wave signal Qm using the latest correction value while correcting the camera characteristics.
Further, the camera characteristic countermeasure process can be configured to end the correction when the correction value converges to a certain extent.
In that case, the correction is continued until the fluctuation due to the camera characteristic becomes at least smaller than the fluctuation of the pulse wave signal.

In this way, the camera characteristic data 15 is completed, and thereafter, by using this, it is not necessary to perform a camera characteristic data update process in pulse wave detection. As a result, the load on the CPU 2 is reduced, and the amount can be used for other processing.
In this case, the pulse wave detection device 1 determines the correction value when the variation in the color space component due to the characteristics of the camera converges to at least a value smaller than the variation in the color space component due to the pulse wave. Complete the update.

According to the third embodiment, the following effects can be obtained.
(1) Camera characteristics can be detected and corrected while using the pulse wave detector 1. Therefore, no prior adjustment is necessary.
(2) Since the correction value is created in accordance with the movement of the user, for example, it is suitable for being mounted on a vehicle and targeting a passenger.
(3) Since the color to be corrected is limited to the face color, complicated calibration calculation is not required, and frame dropping during moving image processing can be suppressed.
(4) As a result of the suppression of frame dropping, the time resolution of the pulse wave is improved. Therefore, it is suitable for real-time processing, and the accuracy of pulse interval (pulse interval) analysis is improved.

Next, the timing for starting the processing (FIGS. 4, 9, and 13) by the pulse wave detection device 1 in the first to third embodiments described will be described.
In each embodiment, the pulse wave detection process by the pulse wave detection device 1 is started at any of the following timings. As each start timing, timing 1 may be set as a default so that the user can change the timing at an arbitrary timing.
(1) Timing 1
The process is started when it is detected that the driver who is the subject of pulse wave monitoring is seated in the driver's seat.
When passengers other than the driver (passenger seats and passengers in the rear seats) are also subject to pulse wave monitoring, the processing is started when it is detected that one of the target passenger seats is detected. .
In this case, for the detection of the passenger, the ignition key is turned on when a load sensor is placed on the target seat (seat surface, backrest portion, etc.) and a load exceeding the threshold is detected, or when the seat belt is worn. If it is (for the driver's seat), it is determined that the user is seated.

(2) Timing 2
A start button is arranged in any of the pulse wave detection devices 1, and processing is started when any of the passengers selects the start button.
In this case, the start button is constituted by the display unit 5 and the input unit 6. That is, the pulse wave detection device 1 displays a start button on the display unit 5 and starts processing when the touch panel of the input unit 6 detects that the corresponding part has been touched.
Further, an independent hard switch may be provided as the input unit 6.
Note that the start button may be provided for each passenger to be monitored.

(3) Timing 3
The process is started when the door of the driver's seat of the vehicle on which the pulse wave detection device 1 is mounted is opened.
If a passenger other than the driver is also a monitoring target, the process is started even when the door corresponding to the corresponding passenger is opened.
The opening / closing of the door is detected by a known technique such as an opening / closing sensor (contact sensor) on the door.
According to this timing 3, it becomes possible to start monitoring the pulse wave earlier than other timings.
In particular, since the process before the pulse wave detection process (step 35) such as the acquisition of skin color data can be completed before the passenger sits after the door is opened, Can be detected.

Next, how to use the detected pulse wave will be described. That is, by using the pulse wave detected by the pulse wave detection device 1 of the present embodiment, it is possible to determine and respond to the driver's state such as sleepiness, tension state, and arousal state such as fatigue.
For example, as a technique for detecting a driver's sleepiness from a pulse wave, there is, for example, Japanese Patent Application Laid-Open No. 2014-20678 “Drowsiness Prediction Device and Sleepiness Prediction System”. As an example, by using this technique, the presence or absence of sleepiness can be monitored from the pulse wave of the driver.
Specifically, the pulse and HF of the driver are measured using the pulse wave detected from the pulse wave detection device 1. HF is a known index that indicates the amount of heartbeat interval fluctuation (heart rate fluctuation).
The driver's sleepiness can be calculated by the following sleepiness value Z.

  Z = P × 10 + (Q−1) × 100

P is a pulse decrease amount (simply bpm) with respect to a normal value, and Q is an increase rate of HF in a predetermined period (for example, the past 500 seconds).
In a state where there is a sign of sleepiness, the sympathetic nerve activity changes from an enhanced state to a suppressed state, and thus the pulse rate decreases. In a state where sleepiness occurs, the parasympathetic nerve changes to an enhanced state, and the pulse rate decreases and HF increases.

The pulse wave detection device 1 monitors a pulse wave, a pulse obtained from the pulse wave, HF, a drowsiness value Z, and the like, and outputs vibration and sound when a drowsiness sign is observed or when drowsiness occurs. , Can alert the driver.
Further, since the pulse changes depending on tension, fatigue, etc. in addition to sleepiness, the pulse wave detection device 1 can monitor the driver's arousal level including these concepts by the pulse wave. It is.

Although the three embodiments have been described above, according to the first embodiment, the following configuration can be obtained.
(1) 101st structure Movie acquisition means for acquiring a moving image obtained by photographing an area including at least the skin of a subject, and a skin part specifying means for specifying a portion of the subject's skin that is reflected in the acquired movie A pulse wave acquisition means for acquiring the pulse wave of the subject from a time change of a predetermined color space component in the specified skin portion, and an output means for outputting the acquired pulse wave. A pulse wave detector.
(2) 102nd Configuration A pulse wave detection device according to a 101st configuration, wherein an area including at least the skin of the subject includes the face of the subject.
(3) 103rd structure It has the reference component registration means which registers the reference component which is the color space component used as the reference | standard for specifying the said subject's skin part, The said skin part specific | specification means is predetermined in the said moving image. The pulse wave detection device according to the 101st configuration or the 102nd configuration, wherein a portion corresponding to the registered reference component is specified as the skin portion.
(4) 104th configuration The color space component used by the skin part specifying unit to specify the skin part and the color space component used by the pulse wave acquiring unit to acquire the pulse wave are different colors. A pulse wave detection device having the 101st configuration, the 102nd configuration, or the 103rd configuration, which is a spatial component.
(5) 105th configuration The color space component used by the skin part specifying means to specify the skin part is an HSV composed of a color component (H), a saturation component (S), and a lightness component (V). The color component (H) of the color space, and the color space component used by the pulse wave acquisition unit to acquire the pulse wave is a YIQ composed of a lightness component (Y) and a chromaticity component (I, Q). The pulse wave detection device according to the 104th configuration, characterized in that it is the chromaticity component (Q) of the color space.
(6) 106th configuration It comprises color space conversion means for converting the color space of a moving image, and the skin part specifying means and the pulse wave acquisition means are converted into a color space obtained by converting the acquired moving image by the color space conversion means. A pulse wave detection device according to any one of the 101st configuration to the 105th configuration, wherein a color space component is acquired.
(7) 107th structure The face image acquisition means which acquires the face image which image | photographed the said subject's face, and the predetermined area | region where the skin of the said face was exposed by face recognition processing in the acquired face image A pulse wave detection device according to a 103rd configuration, wherein the reference component registration unit registers a color space component of the specified region as the reference component.
(8) 108th Configuration The pulse wave detection device according to a 107th configuration, wherein the predetermined region is a region of the subject's nose.
(9) Ninth Configuration The tenth configuration is characterized in that the reference component registration means registers, as the reference component, a value obtained by performing predetermined statistical processing on the distribution of the color space component in the specified area. A pulse wave detection device having the configuration or the 108th configuration.
(10) The 110th configuration Any one of the 101st configuration to the 109th configuration, wherein the skin portion specifying means specifies the skin portion in pixel units of the moving image Pulse wave detector.
(11) Eleventh Configuration The target person is a passenger of a transport device, and comprises a monitoring means for monitoring the physical condition of the passenger using the output pulse wave. A pulse wave detection device having any one of the configurations to the 110th configuration.
(12) The 112th structure The moving image acquisition function which acquires the moving image which image | photographed the area | region containing at least skin of a subject, The skin part specific function which specifies the part of the subject's skin reflected in the acquired moving image, And a pulse wave acquisition function for acquiring a pulse wave of the subject from a time change of a predetermined color space component in the specified skin portion and an output function for outputting the acquired pulse wave by a computer. Wave detection program.

With the above configuration, the following effects can be obtained.
(1) According to the 101st configuration, it is possible to extract only the skin part by excluding the disturbance elements shown in the moving image, and to detect the pulse wave therefrom, so that the accuracy of pulse wave detection can be improved. .
(2) According to the 102nd configuration, it is possible to detect a pulse wave from a face that is exposed daily and that is easy to photograph.
(3) According to the 103rd configuration, it is possible to easily extract a skin portion from a moving image by comparison with a reference component.
(4) According to the 104th configuration, the robustness against disturbance elements is improved by adopting a combination of color space components suitable for the observation target (because the object to be observed with light is different between the skin and the pulse wave). be able to.
(5) According to the 105th structure, a disturbance element can be obtained by combining the H component found to be suitable for specifying the skin part and the Q component found to be suitable for pulse wave detection. The robustness against the can be further improved.
(6) According to the 106th configuration, by providing the color conversion processing not in the external device but in the pulse wave detection device, the processing speed is improved and it becomes easy to detect the pulse wave in real time from the moving image. .
(7) According to the 107th configuration, by collecting a skin color reference component from the subject himself / herself, a reference value including a subtle difference in skin color for each person can be easily obtained. .
(8) According to the 108th configuration, the skin color reference value can be collected from the nose region where the skin is exposed and the location can be easily identified.
(9) According to the 109th configuration, it is possible to equalize the uneven color distribution of the skin having a large individual difference by statistical processing, thereby improving the reliability of the reference component.
(10) According to the 110th configuration, since the skin portion is extracted for each pixel, not the region surrounded by the closed curve (pixels not corresponding to the skin portion are scattered), it acts as a disturbance element from the evaluation target. Pixels can be excluded, and detection accuracy can be increased.
(11) According to the 111th structure, the physical condition of the passenger of a transport equipment can be monitored.
(12) According to the 112th structure, a pulse wave detection apparatus can be easily and inexpensively configured by distributing a pulse wave detection program and installing it in a general-purpose computer.

Although the three embodiments have been described above, according to the second embodiment, the following configurations can be obtained.
(1) 201st structure The moving image acquisition means to acquire the moving image which image | photographed the area | region containing a subject's face, the eye part specification means to specify the eye part of the said subject reflected in the said moving image, and the said moving image A brightness change acquisition means for acquiring a change in brightness caused by a change in the shooting environment from a change in a predetermined color space component of the identified eye portion, and the brightness of the moving image is calculated using the acquired brightness change. Lightness correction means for correcting, pulse wave acquisition means for acquiring the pulse wave of the subject from a time change of a predetermined color space component in the corrected skin portion of the subject, and outputting the acquired pulse wave And a pulse wave detecting device.
(2) No. 202 Configuration comprising a skin part specifying means for specifying the skin part of the subject shown in the acquired moving image, wherein the pulse wave acquiring means is a predetermined color space in the specified skin part 201. A pulse wave detection apparatus having a 201st configuration, wherein the pulse wave of the subject is acquired from a time change of a component.
(3) 203rd structure It has the reference component registration means which registers the reference component which is a color space component used as the reference | standard for specifying the eye part of the said subject, The said eye part specification means is predetermined in the said moving image. The pulse wave detection device according to the 202nd configuration, wherein a portion corresponding to the registered reference component is specified as the eye portion.
(4) 204th Configuration The 202nd skin part specifying means specifies, as the skin part, a part corresponding to a predetermined reference component in which a predetermined color space component is registered in advance in the moving image. A pulse wave detection device having a configuration or a 203rd configuration.
(5) No. 205 configuration Color space component used by the brightness change acquisition means for acquiring a change in brightness, color space component used by the pulse wave acquisition means for acquiring the pulse wave, and the skin portion The pulse wave detection device having the 202nd configuration, the 203rd configuration, or the 204th configuration, wherein the color space component used by the specifying unit to specify the skin portion is a different color space component.
(6) 206th configuration The color space component used for the brightness change acquisition means to acquire a change in brightness and the color space component used for the pulse wave acquisition means to acquire the pulse wave are respectively a brightness component. (Y) and the lightness component (Y) and the chromaticity component (Q) in the YIQ color space comprising the chromaticity components (I, Q), and the skin part specifying means specifies the skin part. The color space component used for is the color component (H) of the HSV color space comprising the color component (H), the saturation component (S), and the lightness component (V). Pulse wave detector.
(7) The 207th configuration It is provided with a color space converting means for converting a color space, and the pulse wave acquiring means, the brightness change acquiring means, and the skin part specifying means are configured to convert the acquired moving image into the color space converting means. A pulse wave detection device according to any one of the 202nd to 206th configurations, wherein a color space component is acquired in the converted color space.
(8) No. 208 configuration Face image acquisition means for acquiring a face image obtained by photographing the face of the subject, and area specification means for specifying the eye area of the face by face recognition processing in the acquired face image And the reference component registration means registers the color space component of the specified area as the reference component, according to the 203rd configuration.
(9) Configuration 209 The reference component registration means registers, as the reference component, a value obtained by performing predetermined statistical processing on the distribution of the color space component in the specified region. Configuration pulse wave detection device.
(10) 210th Configuration The eye portion specifying means specifies any one of the 201 portions to the 207th configuration, wherein the eye portion is specified in pixel units of the moving image. Pulse wave detector.
(11) No. 211 Configuration The pulse wave detection according to any one of the No. 201 to No. 210 configurations, wherein the brightness correction means performs the correction in pixel units of the moving image. apparatus.
(12) Structure No. 212, wherein the subject is a passenger of a transport device, and comprises monitoring means for monitoring the physical condition of the passenger using the output pulse wave. The pulse wave detection apparatus of any one structure from a structure to the 211nd structure.
(13) 213 Configuration A moving image acquisition function for acquiring a moving image in which an area including the face of the subject is photographed, an eye part specifying function for specifying an eye portion of the subject reflected in the moving image, and the moving image A brightness change acquisition function that acquires a change in brightness caused by a change in the shooting environment from a change in a predetermined color space component of the identified eye portion, and the brightness of the moving image is calculated using the acquired brightness change. A brightness correction function to correct, a pulse wave acquisition function to acquire the pulse wave of the subject from a time change of a predetermined color space component in the corrected skin portion of the subject, and output the acquired pulse wave A pulse wave detection program that realizes an output function with a computer.

With the above configuration, the following effects can be obtained.
(1) According to the 201st configuration, it is possible to correct the brightness of a moving image by acquiring a change in brightness from the eye part of the subject.
(2) According to the 202nd configuration, it is possible to extract only a portion of the skin excluding the disturbance elements shown in the moving image, and to detect a pulse wave therefrom, so that the accuracy of pulse wave detection can be improved. .
(3) According to the 203rd configuration, it is possible to take out only the eye portion by excluding the disturbance element shown in the moving image, and improve the lightness correction accuracy.
(4) According to the 204th configuration, a skin portion can be easily extracted from a moving image by comparison with a reference component.
(5) According to the 205th configuration (for lightness change, skin, and pulse wave, the object to be observed with light is different), by adopting a combination of color space components suitable for the observation object, Robustness can be improved.
(6) According to the 206th configuration, the Y component found to be suitable for detecting a change in brightness, the H component found to be suitable for specifying a skin part, and pulse wave detection By combining the Q components that have been found to be suitable, the robustness against disturbance elements can be further improved.
(7) According to the 207th configuration, by providing the color space conversion processing inside the pulse wave detection device instead of the external device, the processing speed is improved and it is easy to detect the pulse wave in real time from the moving image. Become.
(8) According to the 208th structure, it is possible to easily obtain a reference value including a subtle difference in eye color for each person by collecting a reference value of the eye color from the subject himself / herself. it can.
(9) According to the 209th configuration, it is possible to equalize the uneven color distribution of the eyes with a large individual difference by statistical processing, thereby improving the reliability of the reference component.
(10) According to the 210th configuration, an eye part is extracted for each pixel, not an area surrounded by a closed curve (pixels not corresponding to the eye part are also scattered), and thus acts as a disturbance element from the evaluation target. Pixels can be excluded, and detection accuracy can be increased.
(11) According to the 211st configuration, since the area surrounded by the closed curve is not corrected with the representative value but can be corrected for each pixel, the detection accuracy can be improved.
(12) According to the 212th configuration, the physical condition of the passenger of the transport device can be monitored.
(13) According to the 213th configuration, a pulse wave detection device can be easily and inexpensively configured by distributing a pulse wave detection program and installing it in a general-purpose computer.

1 Pulse wave detector 2 CPU
3 ROM
4 RAM
DESCRIPTION OF SYMBOLS 5 Display part 6 Input part 7 Output part 8 Camera 9 Memory | storage part 10 User 12 Pulse wave detection program 14 User database 15 Camera characteristic data 20 Nose area 21 Hair 22 Eyebrow 23 Eye 24 Lip 25 Background 26 Skin part 30 Still image 31 Frame image 32 HSV image 35 Pulse wave 41, 42 Period 45 Eye part 46 White eye part 47 Iris part 48 Pupil part 51 Lightness signal 52 Pre-correction pulse wave signal 53 Post-correction pulse wave signal 55 Eye area 61 Left area 62 Central area 63 Right area 101 Evaluation area 102 Background

Claims (14)

A moving image acquisition means for acquiring a moving image of an area including at least the skin of the subject with a predetermined camera;
A skin part specifying means for specifying the skin part of the subject shown in the acquired video;
Variation correcting means for correcting variation of a predetermined color space component generated in the moving image due to the characteristics of the camera with respect to the pixel of the specified skin portion ;
Pulse wave acquisition means for acquiring the pulse wave of the subject from the temporal change of the color space component in the corrected skin portion;
Output means for outputting the acquired pulse wave;
A pulse wave detection device comprising: The video acquisition unit acquires videos the face of the subject photographed including region,
The variation correction means corrects variations in color space components by limiting the correction target color to the color of the subject's face,
The pulse wave detection device according to claim 1. A moving image acquisition means for acquiring a moving image of an area including at least the skin of the subject with a predetermined camera;
A skin part specifying means for specifying the skin part of the subject shown in the acquired video;
Variation correcting means for correcting variations in a predetermined color space component generated in the moving image due to the characteristics of the camera;
Pulse wave acquisition means for acquiring the pulse wave of the subject from the temporal change of the color space component in the corrected skin portion;
Output means for outputting the acquired pulse wave;
Updating means for updating a correction value used for the correction by applying a predetermined statistical process to a change in the color space component generated in the skin part as the skin part moves;
A pulse wave detection device comprising: A moving image acquisition means for acquiring a moving image of an area including at least the skin of the subject with a predetermined camera;
A skin part specifying means for specifying the skin part of the subject shown in the acquired video;
Variation correcting means for correcting variations in a predetermined color space component generated in the moving image due to the characteristics of the camera;
Pulse wave acquisition means for acquiring the pulse wave of the subject from the temporal change of the color space component in the corrected skin portion;
Output means for outputting the acquired pulse wave ,
The color space component to be corrected by the variation correcting unit and the color space component used by the pulse wave acquiring unit to acquire the pulse wave are a lightness component (Y) and a chromaticity component (I, Q). The chromaticity component (Q) of the YIQ color space consisting of:
A pulse wave detection device characterized by that. An update means for updating a correction value used for the correction by performing a predetermined statistical process on a change in the color space component generated in the skin portion as the skin portion moves is provided. The pulse wave detection device according to claim 1, claim 2, and claim 4 . The update means updates the correction value for each pixel constituting the moving image,
6. The pulse wave detection device according to claim 3, wherein the variation correction unit corrects the color space component using a correction value corresponding to the pixel for each pixel. The update unit is configured to reduce the color space component variation due to the characteristics of the camera to at least a value smaller than a change amount of the color space component due to the pulse wave of the subject. The pulse wave detection device according to claim 3, wherein the update is completed. The variation correction unit corrects the variation using the latest correction value while the update unit is updating the correction value,
The said pulse wave acquisition means acquires the said pulse wave from the part of the said skin correct | amended with the said latest correction value , The claim 6, The claim 6, The claim 7 characterized by the above-mentioned. Pulse wave detector. Brightness change acquisition means for acquiring a change in brightness caused by a change in the shooting environment of the moving image;
Brightness correction means for correcting the brightness of the moving image using the acquired change in brightness;
With
The pulse wave acquiring unit, wherein the brightness correction according to claim claim 1, further from the time change of the color space components in the corrected said portion of skin by means and acquiring the pulse wave of the subject 8 The pulse wave detection device according to claim 1. The pulse wave detection device according to any one of claims 1 to 9 , wherein the region including at least the skin of the subject includes the face of the subject. The subject is a passenger transportation equipment, among the claims 1 to 10, characterized by comprising monitoring means for monitoring the physical condition of the passenger using the pulse wave the output The pulse wave detection device according to claim 1. A video acquisition function for acquiring a video of an area including at least the skin of the subject with a predetermined camera;
A skin part specifying function for specifying the skin part of the subject shown in the acquired video;
A variation correction function that corrects variation of a predetermined color space component that occurs in the moving image due to the characteristics of the camera with respect to the pixels of the identified skin portion ;
A pulse wave acquisition function for acquiring a pulse wave of the subject from a time change of the color space component in the corrected portion of the skin;
An output function for outputting the acquired pulse wave;
Is a pulse wave detection program that realizes this with a computer. A video acquisition function for acquiring a video of an area including at least the skin of the subject with a predetermined camera;
A skin part specifying function for specifying the skin part of the subject shown in the acquired video;
A variation correction function for correcting variations in a predetermined color space component generated in the moving image due to the characteristics of the camera;
A pulse wave acquisition function for acquiring a pulse wave of the subject from a time change of the color space component in the corrected portion of the skin;
An output function for outputting the acquired pulse wave;
An update function for updating a correction value used for the correction by applying a predetermined statistical process to a change in the color space component generated in the skin part as the skin part moves;
Is a pulse wave detection program that realizes this with a computer. A video acquisition function for acquiring a video of an area including at least the skin of the subject with a predetermined camera;
A skin part specifying function for specifying the skin part of the subject shown in the acquired video;
A variation correction function for correcting variations in a predetermined color space component generated in the moving image due to the characteristics of the camera;
A pulse wave acquisition function for acquiring a pulse wave of the subject from a time change of the color space component in the corrected portion of the skin;
An output function for outputting the acquired pulse wave, and a pulse wave detection program for realizing by a computer,
The color space component to be corrected by the variation correction function and the color space component used by the pulse wave acquisition function to acquire the pulse wave are a lightness component (Y) and a chromaticity component (I, Q). The chromaticity component (Q) of the YIQ color space consisting of:
A pulse wave detection program characterized by that.