WO2017051415A1

WO2017051415A1 - A system and method for remotely obtaining physiological parameter of a subject

Info

Publication number: WO2017051415A1
Application number: PCT/IL2016/051047
Authority: WO
Inventors: Dmitry GOLDENBERG; Valeriy BLYUS
Original assignee: Sami Shamoon College Of Engineering (rA); Sensority Ltd
Current assignee: Sami Shamoon College Of Engineering (rA); Sensority Ltd
Priority date: 2015-09-21
Filing date: 2016-09-21
Publication date: 2017-03-30
Anticipated expiration: 2018-03-21

Abstract

The present invention relates to a system for detection of a physiological parameter of an individual, comprising at least one camera for capturing images of said individual from a remote location, and a processing unit for analyzing said captured images in order to extract data that represent physiological parameters relative to said individual, wherein said processing unit is adapted for analyzing photoplethysmogram (PPG) signals that reflects changes in volume within an organ or whole body of said individual that result from fluctuations in the amount of blood or air it contains.

Description

A SYSTEM AND METHOD FOR REMOTELY OBTAINING PHYSIOLOGICAL PARAMETER OF A SUBJECT

Field of the Invention

The present invention relates to the field of imaging systems. More particularly, the invention relates to a system and method for obtaining physiological parameter of a subject by optical means such as a camera that is located remotely from the subject.

Background of the invention

A photoplethysmogram (PPG) is an optically obtained plethysmogram, a volumetric measurement of an organ. In the prior-art, a PPG signal is often obtained by using a pulse oximeter which illuminates the skin and measures changes in light absorption. A conventional pulse oximeter monitors the perfusion of blood to the dermis and subcutaneous tissue of the skin. With each cardiac cycle the heart pumps blood to the periphery. Even though this pressure pulse is somewhat damped by the time it reaches the skin, it is enough to distend the arteries and arterioles in the subcutaneous tissue. If the pulse oximeter is attached without compressing the skin, a pressure pulse can also be seen from the venous plexus, as a small secondary peak. However, in order to use the oximeter a pulse oximeter probe must be physically applied to a person's body, usually to the finger.

It is an object of the present invention to provide a system which is capable of detecting physiological parameters of a subject from a remote location, thus eliminating the need to be physically attached to a person's body.

Other objects and advantages of the invention will become apparent as the description proceeds. Summary of the Invention

The present invention relates to a system for detection of a physiological parameter of an individual, comprising at least one camera (e.g., a color digital camera) for capturing images of said individual from a remote location, and a processing unit for analyzing said captured images in order to extract data that represent physiological parameters relative to said individual, wherein said processing unit is adapted for analyzing photoplethysmogram (PPG) signals that reflects changes in volume within an organ or whole body of said individual that result from fluctuations in the amount of blood or air it contains.

According to an embodiment of the invention, the system further comprises a recording queue module for saving a plurality of frames and a monitoring module adapted for grabbing frame per frame from said recoding queue module.

In another aspect, the present invention relates to a method for_remotely obtaining physiological parameter of an individual person, comprising: a) capturing images of said individual from a remote location by using at least one camera; b) processing said captured images in order to extract data that represent physiological parameters relative to said individual, wherein the processing involves analyzing photoplethysmogram (PPG) signals that reflects changes in volume within an organ or whole body of said individual that result from fluctuations in the amount of blood or air it contains.

According to an embodiment of the invention, the method further comprises: a) Initializing and opening camera stream; b) Initializing a raw signal buffer that is adapted to be populated with the processed data; c) Building queue for recording frames taken from the camera stream; d) Detecting region-of-interest (ROI) in the frames; and e) Capturing data from said detected ROI and starting processing the captured data for the extraction of samples for the creation of a PPG signal.

According to an embodiment of the invention, the detected ROI is the face and the forehead region of the person.

According to an embodiment of the invention, the signal buffer is populated with mean values that represent the forehead region of the person.

According to an embodiment of the invention, the processing of the captured image is done in the Green channel of a RGB color model.

Brief Description of the Drawings

In the drawings:

Fig. 1 schematically illustrates a process of obtaining physiological parameter of a subject in accordance with an embodiment of the present invention;

Fig. 2 schematically illustrates a recording queue process, according to an embodiment of the invention;

Fig. 3 schematically illustrates the operating process of a physiological data monitoring module, according to an embodiment of the invention;

Fig.4 shows an exemplary visual layout of a face and forehead detection;

Fig. 5 shows a detailed view of an array of pixels that represent the forehead region as detected in Fig. 4;

Fig. 6 schematically illustrates an image sampling pipeline for extracting mean value from an array of pixels (in a green channel of a RGB color model based camera) that represents the forehead region of a person; Fig. 7 schematically a signal sampling diagram, according to an embodiment of the present invention;

Fig. 8 is a graph that shows a raw PPG discrete time signal;

Fig. 9 shows Band-Pass filtering on 30 seconds the raw PPG signal;

- Fig. 10 shows comparison between source PPG signal and cubic interpolated signal;

- Fig. 11 shows on separate portion of PPG signal, the effect of applying spline cubic interpolation;

Fig. 12 shows the raw PPG signal (top graph), the middle graph represents all frequencies of signal shown at the top graph, and at the bottom graph, after some zooming, the frequency components of the raw PPG signal is shown; and

Fig. 13 shows frequency components of PPG signal.

Detailed Description of the Invention

The following discussion is intended to provide a brief, general description of a suitable computing environment in which the invention may be implemented. While the invention will be described in the general context of program modules that execute in conjunction with an application program that runs on an operating system on a personal computer, those skilled in the art will recognize that the invention may also be implemented in combination with other computer systems, such as existing surveillance systems, in particular airport security systems or other regions that require large area surveillance to detect, recognize and track suspicious persons or unauthorized intruders.

The program modules describe herein include routines, programs, components, data structures, and other types of structures that perform particular tasks or implement particular data types adapted for analyzing photoplethysmogram (PPG) signals (e.g., as obtained by one or more cameras), which reflects changes in volume within an organ or whole body of a person that result from fluctuations in the amount of blood or air it contains. Moreover, those skilled in the art will appreciate that the invention may be practiced with other computer system configurations while using variety type of cameras, including hand-held devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, minicomputers, mainframe computers, and the like. The invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote memory storage devices.

Embodiments of the invention may be implemented as a computer process (method), a computing system, or as an article of manufacture, such as a computer program product or computer readable media. The computer program product may be a computer storage media readable by a computer system and encoding a computer program of instructions for executing a computer process. The computer program product may also be a propagated signal on a carrier readable by a computing system and encoding a computer program of instructions for executing a computer process.

Unless otherwise indicated, the functions described herein may be performed by executable code and instructions stored in computer readable medium and running on one or more processor-based systems. However, state machines, and/or hardwired electronic circuits can also be utilized. Further, with respect to the example processes described herein, not all the process states need to be reached, nor do the states have to be performed in the illustrated order. Further, certain process states that are illustrated as being serially performed can be performed in parallel. According to an embodiment of the present invention, the process of analyzing PPG signals may consist of individual modules, each of which has its effect on the performance of the process and the execution of the method as a whole. Fig. 1 schematically illustrates a process of obtaining physiological parameter of a subject in accordance with an embodiment of the present invention.

The process may involve the following steps:

Initializing and opening camera stream (step 11);

Initializing a raw signal buffer (step 12);

- Building queue for frames record (step 13);

Detecting region-of-interest (ROI) for capturing data. For example, the ROI can be the face and forehead of a person;

- Upon detection of the ROI, capturing data from the ROI and starting processing the captured data for the extraction of samples of the PPG signal (step 15);

- Monitoring physiological data from the processed data (step 17); and

Recording frames from camera stream (step 16). For example, a monitoring module may grab frame per frame from recorded frames.

One of the most important points of the method is to initialize camera settings and issued images as input to the process which subsequently build PPG signal. As described hereinabove, the process may involve the recognition of the face and forehead of a person, which may serve as an area (i.e., the ROI) for the extraction of samples of the PPG signal.

It is assumed that the brightness of optical signal pulses is relative to the amount of oxygen in the blood. Oxygenated blood is bright red and deoxygenated blood is dark red. From study researches conducted before, it was revealed that part of the forehead of a person is more intense flushing. Accordingly, analyzing PPG signals from monitoring the forehead region of a person yielded the best results, and therefore, the method of monitoring physiological data is more effective when working with forehead region, although other regions of the human body can also be monitored (e.g., by capturing images other ROI of the human body with one or more remote cameras).

The implementation of the monitoring modules takes expensive processor's time, which affects the number of frames processed per second. Therefore, the system may not always have time to process all the frames that the camera is able to produce in one second. To avoid potential loss of signal samples, there is a need to store the frames (preferably to save all the frames that are captured by the camera) in a queue from which they will be processed in next. Fig. 2 schematically illustrates an exemplary recording queue process that may save in a memory all frames that are captured by the camera. The recording queue process may involve the following steps: checking whether the queue is full (step 21), if not full, obtaining a frame from the camera (step 22) and storing the obtained frame in the queue (step 23). As shown in the figure, this process may repeat itself during the entire operating session of the system.

The system may record all possible frames given by the camera and at the same time (parallel) a data monitoring module takes frame per frame from the recording queue and transforms these frames to the signal samples, the conversion processes will be described herein with respect to Fig. 3. The data monitoring module builds and processes PPG signal. The system adds these samples to signal buffer which contains two other samples of the signal during the last few seconds (e.g., last 10 seconds). Since the extracted signal has a lot of noise, defects and unnecessary frequencies, signal must to be processed and/or filtered. According to an embodiment of the invention, the conversion process as performed by the data monitoring module may involve the following steps:

Taking a frame from the queue (step 31). A single frame is indicated by numeral 40 in Figs. 4-7, while a series of frames is indicated by Fl to Fn in Fig. 7;

Acquiring a sample of the forehead of a person (or at least from a partial head view of that person) (step 32). For example, an array of pixels that represents a sample of the forehead of a person that appears in frame 40 is indicated by numeral 42 in Figs. 4-7;

- Acquiring a sample of the background (step 33). For example, a sample of the background is indicated by numeral 43 in Fig. 7;

Subtracting the background sample from the forehead sample as to receive a subtracted value (Sn) (step 34). For example, a mean value of the forehead sample is indicated by numeral 72 in Fig. 7 and a corresponding value of the background sample is indicated by numeral 73 in Fig. 7, while the subtracted value of them is represented by numeral 74 in Fig. 7. In this example, the subtracted value is Sn = 47.93342;

- Adding the subtracted sample to a raw signal buffer (step 35). For example, the raw signal buffer is indicated by numeral 71 in Fig. 7. As indicated in Fig. 7, the signal buffer 71 is populated with substracted values (SI to Sn) over time (as indicated by Tl to Tn); Checking whether the signal buffer is ready (i.e., if the buffer size is above a predetermined value). If the buffer is not ready, repeating steps 31 to 35 until the buffer size will be above or at least equal to the predetermined value (step 36);

If the signal buffer is ready, processing the signal buffer by applying one or more signal processing algorithms to the signal buffer (step 37), such as band-pass filter, median filter and cubic interpolation filter as will be described in further details hereinafter; Applying a Fast Fourier transform (FFT) algorithm to the processed signal buffer as to convert the signal buffer to a representation in the frequency domain (step 38); and

Applying feature extraction to the FFT representation of the signal buffer (step 39).

As aforementioned, the steps of the conversion process described hereinabove are also schematically illustrated with respect to the signal sampling diagram of Fig. 7.

At first, the system applies a band-pass filter to extract the frequency bands of interest. For example, we might select frequencies within. The human heart rate is within 24 - 240 beats per minute, corresponding to 0.4 - 4.0 Hz frequency band (see the term Heart Rate Variability (HRV) at https://en.wikipedia.org/wiki/Heart_rate_variability). Next, the system performs some kind of noise reduction on previously band-passed signal, for example, by applying nonlinear Median filter. Due to the limited camera frame rate, signal has low resolution, so a spline cubic interpolation can be applied to increase signal resolution by interpolating missing samples.

At the end, the system performs Fast Fourier Transform (FFT) to transform the signal into a frequency domain for extracting the heart rate in Beat per Minuets (BPM) and other important physiological features.

According to an embodiment of the invention, an exemplary pseudo code for a single cycle may involve the following tasks:

1. Initialize remote camera.

2. Detect the face of a person.

3. Detect the forehead.

4. Record frames (separate thread) 4.1. Getting camera Frame

4.2. Storing frame to queue

5. Getting frame from queue.

5.1. Get signal sample.

5.1.1. Get Forehead region sample

5.1.2. Get Background region sample

5.1.3. Subtract background sample from forehead sample.

5.2. Add sample to signal buffer

5.3. Process signal.

5.3.1. Bund-pass filter.

5.3.2. Median filter.

5.3.3. Cubic interpolation.

5.4. Perform FFT on signal.

5.4.1. Extract features.

6. Analyze extracted features.

7. Display data.

Remote Camera Initialization

The camera can be any optical instrument capable of recording images, which may be stored locally, transmitted to another location, or both. For example, the camera may include a charge-coupled device (CCD) or other type of sensor to capture images. The resolution of the captured images affects the performance of the algorithm because the image is a key resource, it is well known that the more size (i.e., higher resolution) the more information is necessary to rework algorithm during execution. So the higher the resolution and size of the image the less Frames per Second (FPS) are captured. For example, a relatively low image capture size can be an image of 320 x 240 pixels. Detect (Region of Interest) ROI for signal capture

According to an embodiment of the invention, the system may find and process the face - forehead coordinates of at least one person at one execution time. Before starting the processing of the captured data, the system may continuously monitor and record the coordinates of the face and forehead of the at least one person. When recording starts, signal information may obtained from the last recorded coordinates of the face and forehead. Therefore, it is preferred that a subject person shall not be moving the entire execution process of recording a signal.

Detection of face during the session of signal recording and of data processing, can dramatically reduce the performance of the system. Due to this fact, the coordinates of face - forehead region may be fixed for further work with this ROI.

Face detection

Open source computer vision (OpenCV) library provides fast method of human face detection - Face Detection using Haar Cascades. Object Detection using Haar feature-based cascade classifiers is an effective object detection method proposed by Paul Viola and Michael Jones (herein Viola-Jones method), "Rapid Object Detection using a Boosted Cascade of Simple", Computer Vision and Pattern Recognition, 2001. CVPR 2001. Proceedings of the 2001 IEEE Computer Society Conference on (Volume: 1) 2001.

It is a machine learning based approach where a cascade function is trained from a lot of positive and negative images. It is then used to detect objects in other images.

Open source computer vision (OpenCV), which is a library of programming functions mainly aimed at real-time computer vision, already contains pre-trained classifiers for face detection, so first we need to load the required XML classifiers and do some initialization. Then load our input image in grayscale (e.g., see http://opencv-python- tutroals.readthedocs.io/en/latest/py_tutorials/py_objdetect/py_face_detecti on/py_face_detection . html ).

Before applying Viola-Jones method we have to convert 16-bit / 32-bit RGB frame to 8-bit grayscale frame. The conversion will have some information lose but it will save the execution time of Viola- Jones method.

Suppose the frame represented as RGB color matrix M[i] [j] with i columns and j rows:

Where R, G, and B are integers between 0 and 255. The grayscale weighted average, X for each M[i] [j] is given by the formula:

RGB(M [i] [j])→ Gray: x = 0.299Λ + 0.587G + 0.1145

As the last step before face detection, we apply histogram equalization to improve the contrast in an image. To accomplish the equalization effect the remapping between one distribution (histogram) to another distribution should be the cumulative distribution function. For the histogram H(i), its cumulative distribution H'(i) is:

Finally, we use a simple remapping procedure to obtain the intensity values of the equalized image: exualized(i,j) = H'( [i] [/])

Now we can find the face in the image (see Fig. 4 that shows a frame 40 in which the detected face of a person is indicated by an oval line 41 and the forehead region of this person is indicated by a rectangular line 42). If face is found, it returns the positions of detected faces as Rect(x, y, w, h). From this rectangle we can create ROI on input images for further signal sampling. Fig. 5 shows a detailed view of an array of pixels that represent the forehead region as indicated by the rectangular 42 in Fig. 4. Fig. 6 schematically illustrates an example of image sampling pipeline for extracting a mean value (as indicated by numeral 44) from the array of pixels (e.g., in Green channel of the RGB color model) that represent the forehead region within the boundaries of rectangular 42. In this example, the mean value of the forehead sample is 156.34123.

Forehead detection

Based on the coordinates of the face we can easily calculate the coordinates of the forehead without applying some methods. Received the forehead region, we focus on a small amount of data which is clearly visible pulsation of pixels brightness. For example, see "Remote plethysmographic imaging using ambient light", Wim Verkruysse, et. al. Opt Express. 2008 Dec 22; 16(26): 21434-21445.

Given face coordinates represented as rectangle:

Tect _ace — { Xf_ace, Yface> ^face> Hface} easily can calculate approximated forehead coordinates as rectangle: forehead = w f,ace * 0.25

X forehead ~ (Xface + (Wface * 0.5))— (Wf_orehead

Yforehead ^~ (Jface + ( face * 0.15))— {Hforehead

So we get final forehead coordinates as rectangle:

Frames record queue.

As mentioned earlier, recorder queue captures all frames issued by the camera and if the sampling frequency is less than the camera fps, we do not lose more than one frame because the monitoring unit after the end of the sampling frame refers to recorder queue for sampling the next frame in the queue.

Determine the size Q_Size of the recording Queue for maximum utilization of all possible sampling frames:

Qssii.ze = (7V * FPS) - (7V * _s)

Where T_r is a recording time, FPS is a number of frames per second issued by camera and F_s is the average number of samples obtained in one second, thus F_s = 1/T while T is a constant of time interval between samples. Assuming that FPS > F_s we can get delay time of our sampling system:

T_d is the time required to sample all the remaining frames in the queue, after finished recording signal. So the total time T_t required by system to produce results is:

T_t = T_r + T_d Extracting and sampling signal data

When a new image is obtained from the camera, to the system extracts the region of the forehead and convert it into one single sample and add the sample to the signal. Due to the fact that around the subject's forehead there are other items pulsing at different frequencies (e.g. a lamp or a wall) which can be located in frequency bands of interest, the system may apply some preprocessing to compensate the presence of not needed frequencies. So, in the vision of the time series of samples of the signal, the system may need to subtract the background signal from the forehead signal (as described herein with respect to Figs. 3 and 7).

Sampling forehead region

From the region of forehead (as indicated by the rectangular 42 in Fig. 4), the system extracts the green channel as there is the least noise, and calculates the mean value of all pixels in the green channel. In the end we get a single value - that represents the forehead sample (as shown in Fig. 6 - the forehead region is indicated by numeral 42 and the mean value is indicated by numeral 44). Given forehead ROI F represented by RGB matrix M and P[r, g, b]₍; ₎ are RGB pixels:

P[r, #, b](₀,o) "· P[r, g, b]_{(Q )}

½ =

P[r, g, b] ^■■■ P[r, g, b]_{(i )}

By using only green channel we extract from all pixels P[r, g, b]_{(£ )} # values and getting new matrix from with we calculate the average value of all green values: i J

(i * ;^')

x=0 k=0

Sampling background region

From the original image we calculate the mean value of all pixels around the face region. The resulting mean value describes the overall brightness of the entire region around the face at a given time.

The following is an exemplary code for extracting background region:

def get_background_sample ( self, frame):

face_rect = self. face. get _face()

#convert Ipllmage to numpy array

frame - numpy. asarray(frame[:,:], dtype=numpy .uint8 ) cols = frame.shapefO]

rows - frame. shapefl]

# find pixel outside face region background _j>ixels - []

for i in range(cols):

for j in range(rows):

if self.isInRectd, j, face_rect) == False:

background _j>ixels.append(frame[i, j][l ])

# get mean value (from green channel)

background _j>ixels - numpy.asarray (background _j>ixels, dtype=numpy.uint8) return background _j>ixels.mean()

From the exemplary code above it can be seen that we extract all background pixels into 1 dimensional array background pixels with size N and calculates B_mean -background mean value as:

N

Br, ^ background pixels[x]

x=0

Subtracting in time domain

After extracting the mean value of luminance of pixels in both regions, Fmean and B_mean (Forehead and background), we subtract them from each other to get final mean value Sample_vai_ue of our sample:

Sample_vai_ue— \F_mean— B_mean\

Finally we build signal sample as a pair of Sample_value and Sample_tj_me (time in which a frame entered in the queue):

Samplei = Sample_time, Sample_value) Filling a RAW signal buffer

Buffer represents a window of T seconds and contains the entire sample signals in the last 15 seconds of recording. The static size of buffer is a maximum number of frames that camera can produce in T seconds:

Nbuffer = T * FPS

Buffer will be filled with samples Sample; of the signal. This buffer represents a raw PPG discrete time signal X[N_buffer] (herein PPG signal). Fig. 7 schematically demonstrates the filling of the RAW signal buffer 71 with the subtracted values (SI to Sn) according to the process described hereinabove with respect to Fig. 3. Fig. 8 shows a graphic representation of a raw PPG discrete time signal.

Signal Processing

The acquired PPG signal needs some processing before it will be possible to see and retrieve useful information. We apply a band-pass filter to extract the frequency bands of interest (frequency band of HRV). Then we need to apply noise reduction and interpolate the PPG signal to increase the signal sampling rate.

Band-Pass filtering

According to an embodiment of the invention, the system applies a band pass filter with low-cut 0.7 and high-cut 4.0 to get only the frequencies in bands of interest. The band pass filter helps in both removing the DC component, due to face movement or changes in venous pressure, and also high frequency noise. The result of the band pass filtering (on 30 seconds PPG signal) is shown in Fig. 9. Median filtering— Noise reduction.

In case of small SNR (Signal to noise ratio) value, a median filter or any other nonlinear digital filtering technique to remove noise can be applied to perform some kind of noise reduction from the PPG signal.

Interpolation

One of the biggest issues for HRV analysis is the low resolution of the signal, due to the limited frame rate we can obtain from simple camera. Applying a cubic spline interpolation may increase the number of samples to 50 times (see http://docs.scipy.org/doc/scipy-

0.14.0/reference/tutorial/interpolate.html#id5). This is fast enough to increase the resolution of the signal.

Having a discrete time signal X[n] with knots {(χι, η ) · i = 0, 1, ... , n} we interpolate between all the pairs of knots (x_i-₁, n_i_₁) and (x^ ni) with polynomials: n = qi(x , i = 1, 2, ... , n

The curvature of a curve x = {x[n]} is given

As the spline will take a shape that minimizes the bending both n' and n" will be continuously everywhere and at the knots. To achieve this one must have that:

1 < t < n - 1

<7;'Oi) = <7;+iOi) Fig. 10 shows the resolution issue and the improved signal after cubic spline interpolation. This figure shows the comparison between the source PPG signal and the cubic interpolated signal. Looking at the separate portion of the PPG signal (Fig. 11), it can be seen as a cubic interpolation compensates for the low resolution of the signal.

FFT performing and features extracting

PPG signal gives a visual indication of the heart rate variability but for deeper and more extensive studying of the obtained PPG signal is required to switch from time domain to frequency domain.

A fast Fourier transform (FFT) algorithm allows us to compute the discrete Fourier transform (DFT). FFT converts the signal to the frequency domain and allows you to see in detail all the frequency components of the signal.

Before applying the band-pass filter, you can see a lot of other frequencies of which is RAW signal composed, many of them are the result of the radiation characteristics of foreign objects on the background of the observed subject. Fig. 12 shows the raw PPG signal (top graph), the middle graph represents all frequencies of signal shown at the top graph, and at the bottom graph after some zooming can be seen frequency components of the raw PPG signal. Also, outside the frequency bands of interest can be seen Mayer waves (cyclic changes in arterial blood pressure) at frequency of 0.1Hz.

BPM extracting

In the frequency-domain, HRV is described as the sum of elementary oscillatory components defined by their frequency and amplitude (power). Once the FFT is computed for the current sliding window contents, magnitude bins (Peaks) in the interest band are spotted. Index i, corresponding to the maximum of the power spectrum. To find Index i we used simple max function that detects maximum bin. Now we should convert the Index i into a frequency value F, as follow: i * FPS

^{F =}—

Where N is the size of the extracted signal. Finally we convert F to beats per minute:

BPM = F * 60

In other words we are getting the frequency in bpm that corresponds to the highest peak, as shown in Fig. 13. At the top graph we can see the band-passed PPG signal in bands of 0.7 - 4.0 Hz, and at the bottom graph the frequency components of this signal with a peak at red circle is shown.

The terms, "for example", "e.g.", "optionally", as used hereinabove, were intended to be used to introduce non-limiting examples.

All the above description and examples have been given for the purpose of illustration and are not intended to limit the invention in any way. Many different computer systems, methods of analysis, electronic and optical elements can be employed, all without exceeding the scope of the invention.

Claims

1. A system for detection of a physiological parameter of an individual, comprising at least one camera for capturing images of said individual from a remote location, and a processing unit for analyzing said captured images in order to extract data that represent physiological parameters relative to said individual, wherein said processing unit is adapted for analyzing photoplethysmogram (PPG) signals that reflects changes in volume within an organ or whole body of said individual that result from fluctuations in the amount of blood or air it contains.

2. A system according to claim 1, further comprising a recording queue module for saving a plurality of frames and a monitoring module adapted for grabbing frame per frame from said recoding queue module.

3. A method for_remotely obtaining physiological parameter of an individual person, comprising:

a) capturing images of said individual from a remote location by using at least one camera;

b) processing said captured images in order to extract data that represent physiological parameters relative to said individual, wherein the processing involves analyzing photoplethysmogram (PPG) signals that reflects changes in volume within an organ or whole body of said individual that result from fluctuations in the amount of blood or air it contains.

4. A method according to claim 3, further comprising:

a) Initializing and opening camera stream; b) Initializing a raw signal buffer that is adapted to be populated with the processed data;

c) Building queue for recording frames taken from the camera stream;

d) Detecting region-of-interest (ROI) in the frames; and e) Capturing data from said detected ROI and starting processing the captured data for the extraction of samples for the creation of a PPG signal.

5. A method according to claim 4, wherein the detected ROI is the face and the forehead region of the person.

6. A method according to claim 4, wherein the signal buffer is populated with mean values that represent the forehead region of the person.

7. A method according to claim 3, wherein the processing of the captured image is done in the Green channel of a RGB color model.