A method and device for extracting respiratory angular velocity from video

Abstract

The present invention discloses a method and device for extracting respiratory angular velocity from a video. The method comprises the following steps: S1, obtaining a video image in a to-be-monitored area, and performing image segmentation on the obtained video image so that the entire image is segmented into an M×N matrix; S2, treating each image segmentation block in the matrix as an independent signal monitoring area, obtaining respiratory motion signals in each area, and obtaining respiratory motion acceleration in the X direction and respiratory motion acceleration in the Y direction; S3, calculating the angle between orthogonal respiratory acceleration vectors to obtain respiratory motion angular velocity. The present invention generates a respiratory signal through angular velocity by extracting accelerations in two orthogonal directions, vertical and horizontal, from continuous video images, and calculating the angle between the accelerations in the two orthogonal directions. This method integrates motion information in the orthogonal directions, can more comprehensively describe pixel changes caused by respiratory motion, and thus makes monitoring results more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of non-contact respiratory monitoring, and in particular to a method and device for extracting respiratory angular velocity from a video. Background Art

[0002] Camera-based video sensing technology has been used for contactless vital sign monitoring and health monitoring. This technology uses image and signal processing algorithms to extract physiological signals from a continuous video frame, including heart rate, heart rate variability, respiratory rate, blood oxygen saturation, and blood pressure. Respiration is the most critical and fundamental monitoring indicator, providing a timely indicator of critical illnesses such as physiological deterioration or failure.

[0003] Current video-based respiratory monitoring technologies commonly use pixel translation to infer respiratory motion. This involves using the optical flow method to calculate the acceleration of chest or abdominal pixels in two orthogonal directions, vertical and horizontal, to generate a respiratory motion signal. This signal and respiration rate are then calculated in a single direction (usually the vertical direction). This approach has been widely studied and adopted in product designs. However, this approach suffers from a common drawback: the strength of the respiratory signal detected in a single direction is dependent on the specific posture and position of the subject in the image (sitting, standing, or lying down). For a sitting subject, the vertical component is strongest, while for a lying subject, the horizontal component is stronger. Therefore, monitoring the respiratory signal in a fixed direction (e.g., vertical) cannot accurately monitor the subject in various postures. Furthermore, monitoring the respiratory signal in a single direction fails to fully utilize the motion components generated by pixel movement in all directions, resulting in a loss of motion energy and a reduction in signal quality. Summary of the Invention

[0004] The purpose of the present invention is to provide a method and device for extracting respiratory angular velocity from a video, so as to solve the problems mentioned in the above background technology.

[0005] To achieve the above object, the present invention provides a method for extracting respiratory angular velocity from a video, comprising:

[0006] S1. Acquire a video image in the area to be monitored, and perform image segmentation on the acquired video image so that the entire image is divided into an M×N matrix;

[0007] S2. Treat each image segmentation block in the matrix as an independent signal monitoring area, obtain the respiratory motion signal of each area, analyze the acceleration components generated in the horizontal and vertical orthogonal directions of each respiratory motion signal, and obtain the respiratory motion X-direction acceleration and the respiratory motion Y-direction acceleration;

[0008] S3, calculating the angle between the orthogonal respiratory acceleration vectors based on the acquired respiratory motion X-direction acceleration and the respiratory motion Y-direction acceleration to obtain the respiratory motion angular velocity;

[0009] S4. Concatenate the angular velocities calculated between adjacent frames in each signal monitoring area in the time domain to form an angular velocity signal, process the angular velocity signal to generate a respiratory signal, select high-quality local respiratory signals based on the signal-to-noise ratio, and fuse them into a one-dimensional respiratory signal as the final output;

[0010] S5. Determine whether the posture of the human body in the image is sleeping or sitting based on the angular velocity distribution of the pixel array in each signal monitoring area.

[0011] Furthermore, the video images acquired in the area to be monitored are continuous video images.

[0012] Furthermore, when the video image is divided into image frames, M+1 dividing lines are set in the horizontal direction of the frame, and N+1 dividing lines are set in the vertical direction of the frame.

[0013] Furthermore, the angular velocity signal is processed by cumsum to generate a breathing signal.

[0014] A device for extracting respiratory angular velocity from a video, comprising:

[0015] An image acquisition module is used to collect and obtain video images in the area to be monitored;

[0016] An image segmentation module is used to segment the acquired video image into an M×N matrix so that each image segmentation block serves as a sub-region of the video image for independent signal monitoring;

[0017] a breathing vector calculation module for calculating the acceleration component in each sub-area;

[0018] A respiratory angular velocity calculation module is used to calculate the acceleration vector angle between acceleration components;

[0019] An angular velocity signal generating module is used to calculate and process the angular velocity signal to generate a breathing signal;

[0020] The posture judgment module is used to judge the posture of the human body in the image.

[0021] Furthermore, the respiratory vector calculation module uses an optical flow method with high sensitivity at a sub-pixel level to calculate the acceleration component in each sub-region.

[0022] Furthermore, the device also includes a UI display module for displaying the respiratory signal and human body posture after monitoring is completed.

[0023] The beneficial effects of the present invention compared with the prior art are:

[0024] (1) The present invention generates a respiratory signal through angular velocity by extracting accelerations in two orthogonal directions, vertical and horizontal, from continuous video images, and calculating the angle between the accelerations in the two orthogonal directions. This method integrates the motion information in the orthogonal directions and can more comprehensively describe the pixel changes caused by respiratory motion, making the monitoring results more accurate. At the same time, the calculation of angular velocity can achieve the same monitoring effect in different human postures (sitting or sleeping). From the measurement principle, it solves the dependence of single-direction acceleration monitoring on body posture, and greatly improves the robustness of monitoring to human posture. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a flow chart of the method of the present invention;

[0026] Figure 2 This is a structural block diagram of the device of the present invention. DETAILED DESCRIPTION

[0027] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0028] Example 1:

[0029] like Figure 1-2 As shown, a method for extracting respiratory angular velocity from a video includes:

[0030] S1. Acquire a video image in the area to be monitored, and perform image segmentation on the acquired video image so that the entire image is divided into an M×N matrix; in this embodiment, the video image acquired in the area to be monitored is a continuous video image; in this embodiment, when the video image is subjected to image segmentation, M+1 segmentation lines are set in the horizontal direction of the image, and N+1 segmentation lines are set in the vertical direction of the image; wherein, when the continuous video image is subjected to image segmentation, a continuous image frame signal can be obtained, so that after the respiratory motion signal in the video image is extracted, the signal data between adjacent frames can be connected into a complete respiratory signal, thereby ensuring the monitoring quality of the respiratory signal; and by segmenting the image, the signal data of each area in the image can be accurately obtained, which provides sufficient data support for subsequent respiratory signal calculation and ensures data accuracy;

[0031] S2. Treat each image segmentation block in the matrix as an independent signal monitoring area, obtain the respiratory motion signal of each area, analyze the acceleration components generated in the horizontal and vertical orthogonal directions of each respiratory motion signal, and obtain the respiratory motion X-direction acceleration and the respiratory motion Y-direction acceleration;

[0032] S3, calculating the angle between the orthogonal respiratory acceleration vectors based on the acquired respiratory motion X-direction acceleration and the respiratory motion Y-direction acceleration to obtain the respiratory motion angular velocity;

[0033] S4. Concatenate the angular velocities calculated between adjacent frames in each signal monitoring area in the time domain to form an angular velocity signal, process the angular velocity signal to generate a respiration signal, select high-quality local respiration signals based on the signal-to-noise ratio, and fuse them into a one-dimensional respiration signal as the final output. In this embodiment, the angular velocity signal is processed by cumsum to generate the respiration signal.

[0034] S5. judging, based on the angular velocity distribution of the pixel array in each signal monitoring area, whether the posture of the human body in the image is sleeping or sitting;

[0035] Among them, each image segmentation block divided in the matrix can be used as an independent signal monitoring area, so that the image screen can perform independent signal monitoring in each image segmentation block; and each signal monitoring area can generate corresponding respiratory motion X-direction acceleration and respiratory motion Y-direction acceleration, then each signal monitoring area can set the motion vector of the respiratory motion X-direction acceleration as dx, and the motion vector of the respiratory motion Y-direction acceleration as dy, therefore, the corresponding angular velocity is arctan2(dx, dy), after calculation by the above calculation formula, the angular velocity signal of each signal monitoring area can be obtained, so that the respiratory signal can be generated by processing the angular velocity signal. Since the higher the signal-to-noise ratio, the better the monitoring quality of the respiratory signal, the high-quality local respiratory signal is selected to be fused into a one-dimensional respiratory signal as the final output, which can make the monitoring quality of the respiratory signal higher and the monitoring accuracy of the respiratory volume higher. After the angular velocity signal is processed by cumsum, its motion position information can be restored by multiple accumulation, so that the generated respiratory signal will be more accurate. At the same time, by dividing the video image into several sub-areas, the angular velocity of each area can be monitored separately, and then the angular velocity spatial distribution state of the pixel array in each signal monitoring area can be used to judge the posture of the human body in the image (that is, to judge whether the human body is in a sitting state or a sleeping state), so that the same monitoring effect can be achieved when the human body is in two different postures, namely, sitting or sleeping. From the measurement principle, the dependence of single-direction acceleration monitoring on body posture is solved, and the robustness of monitoring to human posture is greatly improved.

[0036] As a preferred embodiment of the present invention, a device for extracting respiratory angular velocity from a video is also included, comprising:

[0037] An image acquisition module is used to collect and obtain video images in the area to be monitored;

[0038] An image segmentation module is used to segment the acquired video image into an M×N matrix so that each image segmentation block serves as a sub-region of the video image for independent signal monitoring;

[0039] A breathing vector calculation module, used to calculate the acceleration component in each sub-area;

[0040] A respiratory angular velocity calculation module, configured to calculate the acceleration vector angle between acceleration components; in this embodiment, the respiratory vector calculation module uses an optical flow method with sub-pixel-level high sensitivity to calculate the acceleration components in each sub-region;

[0041] An angular velocity signal generating module is used to calculate and process the angular velocity signal to generate a breathing signal;

[0042] The posture judgment module is used to judge the posture of the human body in the image.

[0043] In this embodiment, the device further includes a UI display module for displaying the respiratory signal and body posture after monitoring is completed, so that the monitoring personnel can clearly know the respiratory status, respiratory rate and body posture of the monitored person.

[0044] In summary, the present invention provides a method and device for extracting respiratory angular velocity from video. By extracting accelerations in two orthogonal directions, vertical and horizontal, from continuous video images, and calculating the angle between the accelerations in the two orthogonal directions, a respiratory signal is generated through the angular velocity. This method integrates motion information in the orthogonal directions, can more comprehensively describe the pixel changes caused by respiratory motion, and make the monitoring results more accurate. At the same time, the calculation of angular velocity can achieve the same monitoring effect in different human postures (sitting or sleeping), which solves the dependence of single-direction acceleration monitoring on body posture from the measurement principle, and greatly improves the robustness of monitoring to human posture.

[0045] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic expressions of the above terms should not be understood as necessarily referring to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification.

[0046] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A method for extracting respiratory angular velocity from a video, characterized in that: include: S1. Acquire a video image in the area to be monitored, and perform image segmentation on the acquired video image so that the entire image is divided into an M×N matrix; S2. Treat each image segmentation block in the matrix as an independent signal monitoring area, obtain the respiratory motion signal of each area, analyze the acceleration components generated in the horizontal and vertical orthogonal directions of each respiratory motion signal, and obtain the respiratory motion X-direction acceleration and the respiratory motion Y-direction acceleration; S3, calculating the angle between the orthogonal respiratory acceleration vectors based on the acquired respiratory motion X-direction acceleration and the respiratory motion Y-direction acceleration to obtain the respiratory motion angular velocity; S4. Concatenate the angular velocities calculated between adjacent frames in each signal monitoring area in the time domain to form an angular velocity signal, process the angular velocity signal to generate a respiratory signal, select high-quality local respiratory signals based on the signal-to-noise ratio, and fuse them into a one-dimensional respiratory signal as the final output; S5. Determine whether the posture of the human body in the image is sleeping or sitting based on the angular velocity distribution of the pixel array in each signal monitoring area.

2. A method for extracting respiratory angular velocity from a video according to claim 1, characterized in that: The video images acquired in the area to be monitored are continuous video images.

3. A method for extracting respiratory angular velocity from a video according to claim 2, characterized in that: When the video image is segmented, M+1 segmentation lines are set in the horizontal direction of the image, and N+1 segmentation lines are set in the vertical direction of the image.

4. The method for extracting respiratory angular velocity from a video according to claim 1, wherein: The angular velocity signal is processed by cumsum to generate a breathing signal.

5. A device for extracting respiratory angular velocity from a video, characterized in that: include: An image acquisition module is used to collect and obtain video images in the area to be monitored; An image segmentation module is used to segment the acquired video image into an M×N matrix so that each image segmentation block serves as a sub-region of the video image for independent signal monitoring; a breathing vector calculation module for calculating the acceleration component in each sub-area; A respiratory angular velocity calculation module is used to calculate the acceleration vector angle between acceleration components; An angular velocity signal generating module is used to calculate and process the angular velocity signal to generate a breathing signal; The posture judgment module is used to judge the posture of the human body in the image.

6. The device for extracting respiratory angular velocity from video according to claim 5, wherein: The respiratory vector calculation module calculates the acceleration component in each sub-region by using an optical flow method with high sensitivity at a sub-pixel level.

7. The device for extracting respiratory angular velocity from video according to claim 5, wherein: It also includes a UI display module for displaying the respiratory signal and human posture after monitoring is completed.

Images