JP2022045503A - Vital data measurement device, vital data measurement method and computer program - Google Patents

Abstract

To provide a vital data measurement device, a vital data measurement method and a computer program which can highly accurately detect vital data of a subject in a non-contact manner.SOLUTION: A vital data measurement device comprises: a pulse wave detection unit 111 which detects pulse wave data from an image captured by a camera; a ROI selection unit 112 which calculates the dispersion of pulse wave data detected in a plurality of set ROIs and selects the ROI where the dispersion is the minimum; an image interpolation processing unit 113 which executes image interpolation processing such that a frame rate of the image becomes a frame rate equal to or greater than a threshold; a filter selection unit 114 which selects a filter according to a distance between the camera and the subject; and a movement average calculation unit 115 which calculates movement average of the pulse wave data with the sample number set according to the brightness. The vital data measurement device removes noise with the filter from the pulse wave data, calculates the movement average for the pulse wave data after noise removal, and calculates the pulse rate of the subject on the basis of the pulse wave data subjected to the movement average processing.SELECTED DRAWING: Figure 2A

Description

The present invention relates to a vital data measuring device, a vital data measuring method, and a computer program.

Patent Document 1 discloses a technique in which a subject is imaged with a camera, a plurality of regions of interest are set in the captured image, and a pulse wave of the subject is detected based on a pixel value detected in each region of interest. It has been disclosed. In Patent Document 1, it is possible to acquire vital data such as a pulse wave of a subject without contact.

Japanese Unexamined Patent Publication No. 2015-205050

However, when a subject is imaged with a camera and vital data is measured from the captured image in a non-contact manner, it is difficult to obtain high measurement accuracy because it is greatly affected by the surrounding environment. That is, in order to estimate the pulse wave from the image captured by the camera, a moving image of the subject's face is acquired, and the change in the subject's complexion that is invisible to the human eye is obtained from the acquired moving image. It is detected, and the pulse wave of the subject is detected from this detection result. Therefore, the pulse wave detection accuracy is greatly affected by various conditions such as the movement of the subject at the time of imaging, the frame rate of the camera, the ambient brightness, and the shooting distance.

The present invention has been made in view of the above problems, and an object thereof is a vital data measuring device capable of detecting vital data of a subject with high accuracy without contact and vital data measurement. To provide methods as well as computer programs.

One aspect of the present invention is a vital data measuring device, which is a pulse wave detecting unit that detects pulse wave data from an image of a subject captured by an imaging unit, and a plurality of interests in the image of the subject. The region is set, the dispersion of the pulse wave data detected in each region of interest is calculated, the region of interest that selects the region of interest that minimizes the dispersion, and the frame rate of the image captured by the imaging unit. When the frame rate is less than a predetermined threshold value, the image interpolation processing unit that performs the image interpolation processing so that the frame rate of the image is equal to or higher than the threshold value, the image pickup unit, and the subject. The pulse wave data is extracted from the filter selection unit that selects the noise removal filter according to the distance of, and the region of interest in which the dispersion is minimized in each image having a frame rate equal to or higher than the threshold value, and the pulse wave data is extracted. It is characterized by including a vital data calculation unit that removes noise with the selected noise removal filter and calculates vital data of the subject based on the pulse wave data after noise removal.

One aspect of the present invention is a vital data measuring method, in which a step of detecting pulse wave data from an image of a subject captured by an imaging unit and a plurality of regions of interest are set in the image of the subject. Then, the dispersion of the pulse wave data detected in each region of interest is calculated, the step of selecting the region of interest that minimizes the dispersion, and the frame rate of the image captured by the imaging unit are acquired, and the frame rate is obtained. When is less than a predetermined threshold, a step of interpolating the image so that the frame rate of the image is equal to or higher than the threshold, and a noise removal filter according to the distance between the image pickup unit and the subject are selected. The pulse wave data is extracted from the step and the region of interest in which the dispersion is minimized in each image having a frame rate equal to or higher than the threshold value, and noise is removed from the extracted pulse wave data by the selected noise removal filter. It is characterized by comprising a step of calculating vital data of the subject based on the pulse wave data after noise removal.

One aspect of the present invention is a computer program for operating a computer as the vital data measuring device according to claims 1 to 5.

According to the present invention, it is possible to detect the vital data of a subject in a non-contact manner with high accuracy.

FIG. 1 is a block diagram showing a configuration of a vital data measuring device and a peripheral device thereof according to an embodiment of the present invention. FIG. 2A is a block diagram showing a configuration of a pulse rate calculation server. FIG. 2B is a block diagram showing a configuration of a personal recognition server. FIG. 2C is a block diagram showing a configuration of a data collection server. FIG. 3 is an explanatory diagram showing an ROI set on the face of a subject, where (a) is a cheek, (b) is a forehead, (c) is a vertical rectangle, and (d) is a horizontal rectangle. show. FIG. 4 is a graph showing pulse wave data detected by each ROI shown in FIG. FIG. 5A is a graph showing changes in pulse wave data with the passage of time, and shows a case where the frame rate is small. FIG. 5B is a graph showing changes in pulse wave data with the passage of time, and shows a case where the frame rate is relatively large. FIG. 6 is an explanatory diagram showing a process of calculating a moving average using three pulse wave data. FIG. 7 is a graph showing pulse wave data obtained by calculating the moving average, (a) shows a graph before calculating the moving average, and (b) shows a graph after calculating the moving average. FIG. 8A is a first fractionated diagram of a flowchart showing a processing procedure of the vital data measuring device according to the present embodiment. FIG. 8B is a second sectional view of a flowchart showing a processing procedure of the vital data measuring device according to the present embodiment. FIG. 9 is a block diagram showing a functional configuration of the vital data measuring device according to the present invention.

Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a configuration of a vital data measurement system 1 and its peripheral devices according to an embodiment of the present invention. As shown in FIG. 1, the vital data measurement system 1 according to the present embodiment includes a pulse rate calculation server 11 (vital data measurement device), a personal recognition server 12, and a data collection server 13. The vital data measurement system 1 acquires an image of a person to be inspected (hereinafter referred to as "subject"), and detects vital data such as the pulse rate of the subject based on the acquired image in a non-contact manner. do.

The vital data measurement system 1 is connected to the camera 2 (imaging unit). The camera 2 is installed on, for example, a movable device, a ceiling portion or a wall surface portion of a room in which the subject is present, and captures an image of the subject. The camera 2 may also be used as a camera used for other purposes, such as a camera for crime prevention monitoring.

The vital data measurement system 1 is also connected to the communication device 4 and the Internet 6 via the router 3. The vital data measurement system 1 is connected to the display 5 via the communication device 4. The display 5 is, for example, a smartphone or tablet terminal carried by a subject, a vital data manager, or the like. The vital data measurement system 1 is also connected to the cloud device 7 via the Internet 6.

2A is a block diagram showing the configuration of the pulse rate calculation server 11, FIG. 2B is a block diagram showing the configuration of the personal recognition server 12, and FIG. 2C is a block diagram showing the configuration of the data acquisition server 13.

As shown in FIG. 2A, the pulse rate calculation server 11 includes a pulse wave detection unit 111, an ROI selection unit 112 (interest region selection unit), an image interpolation processing unit 113, a filter selection unit 114, and a moving average calculation unit. It includes 115 and a vital data calculation unit 116.

The pulse wave detection unit 111 detects the pulse wave data of the subject from the color image captured by the camera 2. The green component contained in the color image is detected, and the pulse wave data is detected based on the change in the blood flow of the subject. Focusing on the absorption characteristics of hemoglobin in the blood of the subject, it has a high peak in the green wavelength band (520 to 600 [nm]). Therefore, by detecting the green component, the pulse wave data of the subject can be detected. Since the method of detecting pulse wave data from a color image is a well-known technique, detailed description thereof will be omitted.

The ROI selection unit 112 selects a desired ROI (Region of Interest) from the image of the subject captured by the camera 2. Hereinafter, a method of selecting a desired ROI will be described with reference to FIGS. 3 and 4.

FIG. 3 is an explanatory diagram showing four ROI q1 to q4 set for the face of the subject. In the present embodiment, for example, as shown in FIG. 3A, a rectangular ROIq1 is set on the cheek of the subject P1. As shown in FIG. 3B, a rectangular ROIq2 is set on the forehead of the subject P1. As shown in FIG. 3C, a vertically long rectangular ROIq3 is set on the face of the subject P1. As shown in FIG. 3D, a horizontally long rectangular ROIq4 is set on the face of the subject P1. Then, the pulse wave data is detected based on the images of each ROI q1 to q4.

The cheeks and forehead of the subject often include resting parts, and considering the detection accuracy of pulse wave data and practical aspects, the rectangular ROIq1 including the cheeks and the rectangular shape including the forehead It is better to include ROIq2. In the present embodiment, an example of setting four ROIs q1 to q4 will be described, but the present invention is not limited to this, and ROIs can be set in parts other than the above, or two, three, and so on. Alternatively, it is possible to set five or more ROIs.

The ROI selection unit 112 selects the ROI having the smallest variance of the pulse wave data detected based on the image of each ROI among the ROIs q1 to q4. 4 (a) to 4 (d) are graphs showing pulse wave data detected in each ROI q1 to q4 shown in FIGS. 3 (a) to 3 (d), respectively. As can be understood from the graphs of FIGS. 4A to 4D, the pulse wave data (FIG. 4A) detected by ROIq1 shown in FIG. 3 has a small variance and is detected by ROIq2 to q4. The pulse wave data (FIGS. 4 (b) to (d)) has a relatively large variance. Therefore, the ROI selection unit 112 selects ROIq1 having the smallest variance. In addition to the variance, it is also possible to select the ROI under other conditions, such as selecting the ROI with the smallest peaks of the peaks and valleys.

The image interpolation processing unit 113 shown in FIG. 2 detects the frame rate FR of the image captured by the camera 2. The image interpolation processing unit 113 also determines whether or not the frame rate FR of the image captured by the camera 2 is equal to or greater than a predetermined threshold value FRth. The threshold value FRth is, for example, 50 [FPS]. When the frame rate FR of the image captured by the camera 2 is less than the threshold value FRth, the image interpolation processing unit 113 performs image interpolation processing such as spline interpolation to increase the frame rate.

That is, the frame rate of the image captured by the camera 2 greatly affects the detection accuracy of the pulse wave data. The higher the frame rate, the higher the accuracy of detecting the pulse wave data because the "peaks" and "valleys" generated in the pulse wave data that fluctuates with the passage of time can be detected with high accuracy. Therefore, when the frame rate FR is less than the threshold value FRth, interpolation processing such as spline interpolation is performed to increase the frame rate.

For example, as shown in FIG. 5A, when the frame rate FR of the image P11 of the subject is less than the threshold FRth (for example, FR = 30 [FPS]), and the images a1, a2, ... A5 are acquired. Performs spline interpolation to increase the number of images. Specifically, two interpolated images b1 and b2 are generated between the images a1 and a2. Similarly, interpolated images b3 and b4 are generated between the images a2 and a3. By doing so, the frame rate can be increased.

On the other hand, when the frame rate FR of the image captured by the camera 2 is equal to or higher than the threshold value FRth, the image interpolation processing unit 113 adopts an image having the same frame rate without performing spline interpolation.

For example, as shown in FIG. 5B, when the frame rate FR of the image P11 of the subject is equal to or higher than the threshold value FRth (for example, FR = 240 [FPS]) and the images a1 to an are acquired, spline interpolation is performed. No processing is performed, and the frame rate is used as it is.

Further, the number of images to be interpolated (referred to as "y") can be calculated by the following equations (1) and (2) when the measurement time is tα.

y = (FRth－FR) * tα (FR <FRth)… (1)
y = 0 (FR ≧ FRth)… (2)
For example, when FR = 30 and FRth = 50, 20 (50-30 = 20) images are interpolated per unit time, and since the measurement time is tα, a total of “20 * tα” images are interpolated. It may be generated by processing. Further, in the case of "FR ≥ FRth", the interpolation processing is not required, so "y = 0" is set.

In addition to spline interpolation, other interpolation methods such as Lagrange interpolation can be used for image interpolation processing. Further, the threshold value FRth of the frame rate FR is not limited to 50 [FPS], and can be appropriately changed according to the performance of the camera 2 and the like. Further, a plurality of threshold values, for example, a first threshold value FRth1 and a second threshold value FRth2 (> FRth1) are set, and the three threshold values are "FR <FRth1", "FRth1≤FR <FRth2", and "FRth2 <FR". It may be divided and the number of images to be interpolated may be set.

The filter selection unit 114 shown in FIG. 2 performs a process of selecting a filter (noise reduction filter) for removing noise included in the pulse wave data according to the distance from the camera 2 to the subject P1.

The longer the distance from the camera 2 to the subject, the more easily the detected pulse wave data is affected by the noise of the high frequency component. Therefore, when the distance from the camera 2 to the subject is equal to or greater than a predetermined reference distance, SGF (Savizky Golay Filter) or BPF (Band), which is effective for removing noise of high frequency components, is effective as a noise reduction filter. Select Pass Filter).

Further, the shorter the distance from the camera 2 to the subject, the more clearly the pulse wave data tends to appear. Therefore, when the distance from the camera 2 to the subject is less than a predetermined reference distance, an ICA (Independent Component Analysis) filter is selected.

In the above-mentioned example, an example in which a reference distance is set and a filter is selected according to whether or not the distance from the camera 2 to the subject is equal to or greater than the reference distance has been described. It may be classified into 3 or more categories such as "30 to 50 cm" and "50 cm or more".

Further, a filter other than the above-mentioned SGF, BPF, and ICA may be used. The distance from the camera 2 to the subject can also be calculated by a depth sensor, a stereo camera, estimation from an arbitrary object, or the like.

It is also possible to perform a Fourier transform process on the obtained pulse wave data to obtain a frequency distribution and select a filter based on the ratio (σ) of the high frequency component contained in the pulse wave data. .. For example, it is possible to select SGF when “σ ≧ 0.3” and set to adopt ICA when “σ <0.3”.

The moving average calculation unit 115 shown in FIG. 2 detects the brightness indicating the brightness of the surroundings of the subject, and sets the number of samples for performing the moving average process according to the detected brightness. The moving average calculation unit 115 can acquire the brightness (v) by converting the image captured by the camera 2 into the HSV format. Alternatively, the brightness may be detected by using a sensor (not shown).

FIG. 6 is an explanatory diagram showing an example of calculating a moving average using three adjacent pulse wave data (that is, the number of samples is “3”). The data d1 to d4 shown in FIG. 6 show the pulse wave data detected in time series, and the data e1 and e2 show the moving average. The moving average e1 is the average value of the three data d1, d2, and d3. The moving average e2 is the average value of the three data d2, d3, and d4. By doing so, the change of the pulse wave data can be smoothed. Further, by increasing the number of samples when calculating the moving average (for example, by setting the number of samples to "5"), the change of the pulse wave data can be made smoother.

FIG. 7A is a graph showing pulse wave data, and FIG. 7B is a graph showing time-series data to which moving average processing is applied. As can be understood from FIGS. 7 (a) and 7 (b), the pulse wave data can be converted into a smooth graph by calculating the moving average.

Then, the moving average calculation unit 115 changes the number of samples for calculating the moving average according to the brightness in the vicinity of the subject. For example, if the reference brightness is set and the brightness around the subject is less than the reference brightness, the number of samples is set to "3", and if the brightness is equal to or higher than the reference brightness, the number of samples is set to "5". The brighter the surroundings of the subject, the more sensitively the reflection of light can be detected. Therefore, the higher the brightness of the surroundings, the larger the number of samples for calculating the moving average. By doing so, it is possible to always obtain stable pulse wave data even when the brightness around the subject changes.

In the above example, an example in which one of the two sample numbers is selected depending on whether or not the brightness is equal to or higher than the standard brightness is shown, but the present invention is not limited to this, and the present invention is not limited to this. It may be classified into the number of samples. The higher the brightness of the surrounding environment, the larger the number of samples at the time of calculating the moving average may be set.

The vital data calculation unit 116 shown in FIG. 2 adds an image interpolation process to the pulse wave data obtained by the above-mentioned processing, that is, the image data of the selected ROI among the plurality of ROI q1 to q4, and the filter selection unit 114. The pulse number of the subject is calculated based on the pulse wave data in which the noise is removed by the filter selected in and the moving average processing is performed by the moving average calculation unit. Since the method of calculating the pulse rate from the pulse wave data is a well-known technique, the description thereof will be omitted.

Next, the personal recognition server 12 shown in FIG. 2B will be described. The personal recognition server 12 includes a user authentication unit 121 and a personal recognition unit 122.

The user authentication unit 121 identifies the ID of the subject. For example, the subject ID is specified from the facial images of the plurality of subjects captured by the camera 2.

The personal recognition unit 122 recognizes a subject to be detected for pulse wave data based on the subject ID specified by the user authentication unit 121.

Next, the data collection server 13 shown in FIG. 2C will be described. The data collection server 13 includes a time-series data extraction unit 131 and a data collection unit 132.

The data collection unit 132 collects image data captured by the camera 2.

The time-series data extraction unit 131 extracts the pulse wave data detected by the pulse wave detection unit 111 from the image data collected by the data collection unit 132, and stores the extracted time-series data. Based on the face image of the subject captured by the camera 2, for example, face recognition processing is performed to recognize the ID of the subject, and the vital data for each subject is stored in the time-series data extraction unit 131. can do. In addition to the face image of the subject, it is also possible to recognize the ID of the subject by adding a QR code (registered trademark) or color barcode to the subject and reading these codes. be. By storing and storing the vital data of each subject, the transition of the vital data of each subject can be viewed.

It is also possible to configure the pulse rate calculation server 11, the personal recognition server 12, and the data collection server 13 shown in FIG. 2A in the cloud device 7 shown in FIG.

[Operation of this embodiment]
Next, the processing procedure of the vital data measurement server 1 according to the present embodiment will be described with reference to the flowcharts shown in FIGS. 8A and 8B.

First, in step S11 shown in FIG. 8A, the image pickup of the color moving image of the subject by the camera 2 is started. The pulse rate calculation server 11 acquires a color moving image captured by the camera 2. Specifically, a color moving image captured by a camera 2 installed in a living room or the like where the subject P1 is located is acquired.

In step S12, the pulse wave detection unit 111 initializes the image number (referred to as “j”) of the image acquired from the camera 2 in time series. That is, "j = 1".

In step S13, the pulse wave detection unit 111 determines whether or not the image number j is larger than the numerical value (FR * t) obtained by multiplying the frame rate FR of the camera 2 by the elapsed time t. That is, it is determined whether or not "j> (FR * t)". If "j> (FR * t)" (S13; YES), the process proceeds to step S18, otherwise (S13; NO), the process proceeds to step S14.

In this processing, for example, when the frame rate (FPS) is "30" and the imaging time is 20 seconds, the right side is 300 * 20 = 600, and processing is performed on 600 images. Therefore, until "j = 600", the process of step S13 is determined as NO.

In step S14, the pulse wave detection unit 111 acquires the image of the image number j.

In step S15, the pulse wave detection unit 111 extracts a plurality of rectangular ROIs (see FIG. 3) set on the face of the subject.

In step S16, the pulse wave detection unit 111 extracts the green component from the image of each ROI and calculates the average value of the green component in the ROI.

In step S17, the pulse wave detection unit 111 increments the image number j (j = j + 1) and returns the process to step S13.

If it is determined in step S13 that "j> (FR * t)", in step S18, the pulse wave detection unit 111 uses the acquired green component as pulse wave data.

In step S19, the pulse wave detection unit 111 calculates the variance of the pulse wave data in each ROI (see FIG. 3) set on the face of the subject. The pulse wave detection unit 111 calculates the variance of the pulse wave data based on the average value of the green components of the pulse wave data described above.

In step S20, the ROI selection unit 112 selects the ROI having the smallest dispersion among the plurality of ROIs.

In step S21 shown in FIG. 8B, the image interpolation processing unit 113 determines whether or not the frame rate FR of the image acquired from the camera 2 is less than a predetermined threshold value FRth. If "FR <FRth" (S21; YES), the process proceeds to step S22, otherwise (S21; NO), the process proceeds to step S23.

In step S22, the image interpolation processing unit 113 increases the frame rate by performing spline interpolation. That is, as described with reference to FIG. 5A described above, when “FR <FRth”, the frame rate is increased by interpolation processing such as spline interpolation. Specifically, the number of "(FRth-FR) * tα" images is increased by interpolation processing. “Tα” is the measurement time.

In step S23, the filter selection unit 114 applies a Fourier transform to the pulse wave data to calculate the ratio σ of the high frequency component of the pulse wave data.

In step S24, the filter selection unit 114 determines whether or not the ratio σ of the high frequency component is equal to or higher than the reference value (for example, 0.3). When the distance from the camera 2 to the subject is less than the reference distance and the ratio σ of the high frequency component is “σ <0.3” (S24; YES), the process proceeds to step S25. If the distance from the camera 2 to the subject is equal to or greater than the reference distance and the ratio σ of the high frequency component is “σ ≧ 0.3” (S24; NO), the process proceeds to step S26. It is also possible to determine S24 based on either the distance from the camera 2 to the subject or the ratio σ of the high frequency component.

In step S25, the filter selection unit 114 selects ICA as a filter for removing noise from the pulse wave data. By selecting ICA, pulse wave data can be acquired with high accuracy.

In step S26, the filter selection unit 114 selects SGF or BPF as a filter for removing noise from the pulse wave data. By selecting SGF or BPF, it is possible to efficiently remove high frequency components.

In step S27, the moving average calculation unit 115 converts the image captured by the camera 2 into the HSV format. That is, it is converted into color space data consisting of three components of hue (Hue), saturation (Saturation), and lightness (Value).

In step S28, the moving average calculation unit 115 determines whether or not the brightness (v) obtained in the process of step S27 is less than a predetermined reference value (for example, 100). If "v <100" (S28; YES), the process proceeds to step S29, and if not (S28; NO), the process proceeds to step S30.

In step S29, the moving average calculation unit 115 sets the number of samples for calculating the moving average to "3".

In step S30, the moving average calculation unit 115 sets the number of samples for calculating the moving average to "5".

In step S31, the vital data calculation unit 116 adds a moving average process to the pulse wave data after noise reduction in step S25 or S26 using the set number of samples. Further, the pulse rate of the subject is calculated based on the pulse wave data to which the moving average processing is added.

In this way, the pulse wave data can be detected with high accuracy, and the vital data of the subject can be calculated with high accuracy.

[Effect of this embodiment]
The vital data measuring device (pulse rate calculation server 11) according to the present embodiment includes a pulse wave detection unit 111 that detects pulse wave data from an image of a subject imaged by an image pickup unit (camera 2) and a test subject. A plurality of ROIs (regions of interest) are set in the image of the person, the dispersion of the pulse wave data detected by each ROI is calculated, and the ROI selection unit 112 that selects the ROI that minimizes the dispersion and the image pickup unit capture the image. The image interpolation processing unit 113 and the imaging unit 113 that acquire the frame rate of the image and perform the image interpolation processing so that the frame rate FR of the image is equal to or more than the threshold FRth when the frame rate FR is less than the predetermined threshold FRth. Pulse wave data is extracted from the filter selection unit 114 that selects the noise removal filter according to the distance between the subject and the subject, and the ROI that minimizes the dispersion of each image with a frame rate higher than the threshold value, and the extracted pulse wave. It is provided with a vital data calculation unit 116 that removes noise with a noise removal filter selected from the data and calculates vital data (pulse rate) of the subject based on the pulse wave data after noise removal.

As described above, in the pulse rate calculation server 11 according to the present embodiment, a plurality of ROIs are set on the face of the subject, and the ROI having the smallest variance of the pulse wave data detected by each ROI is selected. Therefore, the pulse rate can be calculated using the pulse wave data with smaller fluctuations, and the pulse rate can be calculated with high accuracy.

That is, the pulse wave data, which is human body information, is easily affected by noise. Therefore, in order to detect the pulse wave data more accurately, it is preferable to set a plurality of ROIs on the face of the subject, capture an image, and select the ROI that can be taken most stably. In the present embodiment, the detection accuracy of pulse wave data can be improved by setting the ROI of the cheek, the forehead, the vertical rectangle, and the horizontal rectangle with little movement.

Further, in the present embodiment, when the frame rate FR of the image captured by the camera 2 is less than the threshold value FRth, the frame rate is increased by interpolation processing such as spline interpolation. Therefore, it is possible to acquire the data of "peaks" and "valleys" caused by the fluctuation of the pulse wave data, detect the pulse wave data with high accuracy, and improve the detection accuracy of the pulse rate.

Further, in the present embodiment, the larger the brightness around the subject, the larger the number of samples for the moving average processing of the pulse wave data. Therefore, even when the pulse wave data changes sensitively due to the reflection of light, it is possible to correct the change so that the pulse wave data changes smoothly.

Further, in the present embodiment, the filter used to remove the noise of the pulse wave data is changed according to the distance between the subject and the camera 2. Specifically, the longer the distance, the larger the high frequency component σ, so when “σ <0.3”, an ICA filter is used. When “σ ≧ 0.3”, an SGF filter is adopted. Therefore, it is possible to efficiently remove the noise generated by the change in the distance and improve the detection accuracy of the pulse rate. In this embodiment, the subject's movement correction (S19, S20), camera frame rate correction (S21, S22), shooting distance correction (S23 to S26), and ambient brightness correction (S27 to S30) Although the example in which the treatments are carried out in the order of the above is described, the present invention is not limited to the above-mentioned order, and the order of each of the above-mentioned treatments may be interchanged.

In the present embodiment, the pulse wave / pulse rate has been described as an example of the vital data of the subject, but the present invention is not limited to this, and the vital data other than the pulse rate shall be measured. Is also possible.

As shown in FIG. 9, the vital data measuring device (pulse rate calculation server 11) of the present embodiment described above includes, for example, a CPU (Central Processing Unit, processor) 901, a memory 902, and a storage 903 (HDD: Hard). A general-purpose computer system including a Disk Drive (SSD: Solid State Drive), a communication device 904, an input device 905, and an output device 906 can be used. The memory 902 and the storage 903 are storage devices. In this computer system, each function of the vital data measuring device is realized by the CPU 901 executing a predetermined program loaded on the memory 902.

The vital data measuring device may be mounted on one computer or may be mounted on a plurality of computers. Further, the vital data measuring device may be a virtual machine mounted on a computer.

The program for the vital data measuring device can be stored in a computer-readable recording medium such as an HDD, SSD, USB (Universal Serial Bus) memory, CD (Compact Disc), or DVD (Digital Versatile Disc). It can also be delivered via.

The present invention is not limited to the above embodiment, and many modifications can be made within the scope of the gist thereof.

1 Vital data measurement system 2 Camera (imaging unit)
3 Router 4 Communication device 5 Display 6 Internet 7 Cloud device 11 Pulse rate calculation server (vital data measurement device)
12 Personal recognition server 13 Data collection server 111 Pulse wave detection unit 112 ROI selection unit (region of interest selection unit)
113 Image interpolation processing unit 114 Filter selection unit 115 Moving average calculation unit 116 Vital data calculation unit 121 User authentication unit 122 Personal recognition unit 131 Time series data extraction unit 132 Data collection unit

Claims (7)

A pulse wave detection unit that detects pulse wave data from the image of the subject captured by the image pickup unit, and a pulse wave detection unit.
A region of interest selection unit that sets a plurality of regions of interest in the image of the subject, calculates the variance of the pulse wave data detected in each region of interest, and selects the region of interest that minimizes the variance.
Image interpolation processing that acquires the frame rate of the image captured by the image pickup unit and performs image interpolation processing so that the frame rate of the image is equal to or higher than the threshold value when the frame rate is less than a predetermined threshold value. Department and
A filter selection unit that selects a noise reduction filter according to the distance between the image pickup unit and the subject, and a filter selection unit.
After removing noise by extracting pulse wave data from the region of interest in which the dispersion of each image having a frame rate equal to or higher than the threshold value is minimized, and removing noise from the extracted pulse wave data with the selected noise reduction filter. Vital data calculation unit that calculates the vital data of the subject based on the pulse wave data of
Vital data measuring device characterized by being equipped with. A moving average for calculating the moving average of the pulse wave data with respect to the passage of time is set according to the brightness of the surroundings of the subject, and the moving average of the pulse wave data is calculated with the set number of samples. Further equipped with a calculation unit,
The vital data measuring device according to claim 1, wherein the vital data calculation unit calculates vital data of the subject based on the moving average calculated by the moving average calculation unit. The vital data measuring device according to claim 2, wherein the moving average calculation unit sets a larger number of samples for calculating the moving average as the brightness around the subject increases. The vital data measuring device according to any one of claims 1 to 3, wherein the plurality of areas of interest include the cheeks and the forehead of the subject. The filter selection unit according to any one of claims 1 to 4, wherein the filter selection unit selects an SGF filter or a BPF filter when the distance between the image pickup unit and the subject is longer than a predetermined reference distance. The described vital data measuring device. The step of detecting pulse wave data from the image of the subject captured by the image pickup unit,
A step of setting a plurality of regions of interest in the image of the subject, calculating the variance of the pulse wave data detected in each region of interest, and selecting the region of interest that minimizes the variance.
A step of acquiring a frame rate of an image captured by the imaging unit and interpolating the image so that the frame rate of the image is equal to or higher than the threshold value when the frame rate is less than a predetermined threshold value.
A step of selecting a noise reduction filter according to the distance between the image pickup unit and the subject, and
The pulse wave data is extracted from the region of interest in which the dispersion is minimized in each image having a frame rate equal to or higher than the threshold value, and noise is removed from the extracted pulse wave data by the selected noise reduction filter to remove the noise. The step of calculating the vital data of the subject based on the subsequent pulse wave data, and
A vital data measurement method characterized by being equipped with. A computer program that causes a computer to function as the vital data measuring device according to any one of claims 1 to 5.

Images