WO2023236989A1 - 对象跌倒检测方法、装置、毫米波雷达和存储介质 - Google Patents

对象跌倒检测方法、装置、毫米波雷达和存储介质 Download PDF

Info

Publication number
WO2023236989A1
WO2023236989A1 PCT/CN2023/098845 CN2023098845W WO2023236989A1 WO 2023236989 A1 WO2023236989 A1 WO 2023236989A1 CN 2023098845 W CN2023098845 W CN 2023098845W WO 2023236989 A1 WO2023236989 A1 WO 2023236989A1
Authority
WO
WIPO (PCT)
Prior art keywords
point cloud
cloud data
target object
signal strength
signal
Prior art date
Application number
PCT/CN2023/098845
Other languages
English (en)
French (fr)
Inventor
刘百超
刘军辉
张理斌
唐德琴
Original Assignee
长沙莫之比智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 长沙莫之比智能科技有限公司 filed Critical 长沙莫之比智能科技有限公司
Publication of WO2023236989A1 publication Critical patent/WO2023236989A1/zh

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/103—Detecting, measuring or recording devices for testing the shape, pattern, colour, size or movement of the body or parts thereof, for diagnostic purposes
    • A61B5/11—Measuring movement of the entire body or parts thereof, e.g. head or hand tremor, mobility of a limb
    • A61B5/1116—Determining posture transitions
    • A61B5/1117—Fall detection
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7253—Details of waveform analysis characterised by using transforms
    • A61B5/7257—Details of waveform analysis characterised by using transforms using Fourier transforms
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section

Definitions

  • the present application relates to the technical field of fall detection, and in particular to an object fall detection method, device, millimeter-wave radar, computer-readable storage media and computer program products.
  • the target object wears a wearable sensing device, and the sensors in the device sense the speed or acceleration of the human body in multiple directions in the three-dimensional space to determine the fall.
  • the target subject may not wear the wearable sensing device when washing and other prone to falls, and effective detection cannot be performed.
  • Fall detection due to the constraints of the fall judgment algorithm, actions with fall characteristics such as shaking hands can easily be misjudged as falling behaviors. Therefore, traditional object fall detection methods have the disadvantage of poor accuracy of detection results.
  • an object fall detection method is provided to improve the accuracy of object fall detection results.
  • An object fall detection method including:
  • the above object fall detection method can identify the point cloud data of the target object by performing Fourier transform processing and cluster analysis on the echo signal, perform signal strength analysis on the point cloud data, and determine the point cloud corresponding to the point cloud data.
  • the center of mass position of the cluster can eliminate interference from other objects in the environment other than the target object, and the entire fall detection process does not require the target object to wear any auxiliary sensing device. It has strong environmental adaptability and is conducive to improving the accuracy of the target object's fall detection results.
  • signal strength analysis is performed on the point cloud data to determine the centroid position of the point cloud cluster corresponding to the point cloud data, including:
  • the position information and signal strength information in the point cloud data perform signal strength analysis on the point cloud data to obtain the respective position and signal strength of each point cloud in the point cloud data; and based on the respective position and signal strength of each point cloud, determine The centroid position of the point cloud cluster composed of each point cloud; the signal strength is characterized by the signal-to-noise ratio.
  • the position of the center of mass is determined by combining the position information and the signal strength information, which can reduce the impact of noise information and help improve the accuracy of the center of mass position, thereby improving the accuracy of fall detection results.
  • based on the location information and signal strength information in the point cloud data based on the respective position of each point cloud position and signal strength to determine the centroid position of the point cloud cluster composed of each point cloud, including:
  • each point cloud Based on the respective signal strengths of each point cloud, determine the respective weights of each point cloud; and based on the respective weights of each point cloud, perform a weighted sum of the respective positions of each point cloud to determine the centroid of the point cloud cluster composed of each point cloud. Location.
  • centroid of the point cloud cluster is determined based on the signal strength of each point cloud.
  • the algorithm is simple and is conducive to improving data processing efficiency.
  • the respective weights of each point cloud are determined based on the corresponding signal strength information of each point cloud, including:
  • the total energy of the point cloud cluster composed of each point cloud is determined; and for each point cloud, the ratio of the signal strength of the point cloud to the total energy is determined as the weight of the point cloud.
  • the weight of the point cloud data is determined based on the energy distribution of each point cloud data, which can automatically filter out point clouds with smaller energy, which is beneficial to improving the accuracy of the position information of the center of mass.
  • Fourier transform processing and cluster analysis are performed on the echo signals to obtain point cloud data of the target object, including:
  • Fourier transform processing and cluster analysis are performed on the echo signal to obtain point cloud data of the target object.
  • At least one of the attribute characteristics or motion characteristics of the target object is comprehensively considered to obtain the point cloud data of the target object, which is beneficial to improving the matching degree between the point cloud data and the target object, thereby improving the accuracy of fall detection results. sex.
  • Fourier transform processing and cluster analysis are performed on the echo signal in combination with at least one of the attribute characteristics or motion characteristics of the target object in the object to obtain point cloud data of the target object, including:
  • At least one method is to perform cluster analysis on point cloud data containing location information and signal strength information to obtain point cloud data of the target object.
  • the point cloud data of the target object is obtained by performing fast Fourier transform processing and cluster analysis on the echo signal, so as to reduce the amount of calculation and improve the data processing efficiency.
  • the feature point is the origin of radar coordinates; the calculation formula of the feature angle is:
  • ⁇ is the characteristic angle
  • x, y and z jointly represent the position of the center of mass.
  • the characteristic point is defined as the origin of the radar coordinates. Since the position information determined based on the echo signal of the millimeter wave radar corresponds to the coordinate system of the origin of the radar coordinates, there is no need to perform coordinate calculations in the process of calculating the characteristic angle. Conversion can simplify the calculation process of feature angles and help improve data processing efficiency.
  • determining the fall detection result of the target object based on changes in the characteristic angle includes:
  • the fall detection result of the target object is determined to be a fall;
  • the fall determination condition is the difference between each feature angle and the historical feature angle of the previous frame.
  • the proportion of positive or negative values in the values is greater than the set proportion, or the change in the difference between each difference value relative to the previous difference value increases.
  • the fall determination is performed based on the characteristic angles calculated within a set number of consecutive time frames, which is beneficial to improving the accuracy of the fall detection results.
  • determining that the fall detection result of the target object is that a fall has occurred includes:
  • An object fall detection device including:
  • the echo signal acquisition module is used to acquire the echo signal formed after the electromagnetic wave emitted by the millimeter wave radar is reflected by the object;
  • the target object echo signal analysis module is used to perform Fourier transform processing and cluster analysis on the echo signal to obtain point cloud data of the target object;
  • the centroid position determination module is used to perform signal strength analysis on the point cloud data based on the position information and signal strength information in the point cloud data, and determine the centroid position of the point cloud cluster corresponding to the point cloud data;
  • a feature angle calculation module for calculating the feature angle of the centroid relative to the feature point based on the centroid position
  • the fall detection result determination module is used to determine the fall detection result of the target object based on the change of the characteristic angle.
  • a computer device includes a memory and one or more processors.
  • Computer-readable instructions are stored in the memory. When the computer-readable instructions are executed by one or more processors, they cause one or more processors to perform the following steps:
  • One or more non-volatile computer-readable storage media storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the following steps:
  • a computer program product includes a computer program that, when executed by one or more processors, causes one or more processors to perform the following steps:
  • the above-mentioned fall detection device, computer equipment, computer-readable storage medium and computer program product can identify the point cloud data of the target object by performing Fourier transform processing and cluster analysis on the echo signal, and perform signal processing on the point cloud data. Intensity analysis determines the centroid position of the point cloud cluster corresponding to the point cloud data, which can eliminate interference from other objects in the environment other than the target object, and the entire fall detection process does not require the target object to wear any auxiliary sensing devices. It has strong environmental adaptability and has It is beneficial to improve the accuracy of fall detection results of target objects.
  • Figure 1 is an application environment diagram of the object fall detection method in some embodiments of the present application.
  • Figure 2 is a schematic flowchart of an object fall detection method in some embodiments of the present application.
  • Figure 3 is a schematic diagram of the relative positions of feature points and centroids in some embodiments of the present application.
  • Figure 4 is a schematic diagram of the relative positions of feature points and centroids in other embodiments of the present application.
  • Figure 5 is a schematic flowchart of an object fall detection method in other embodiments of the present application.
  • Figure 6 is a schematic diagram of the object fall detection process in some embodiments of the present application.
  • Figure 7 is a schematic flowchart of a fall determination algorithm in some embodiments of the present application.
  • Figure 8 is a structural block diagram of an object fall detection device in some embodiments of the present application.
  • Figure 9 is an internal structural diagram of a millimeter wave radar in some embodiments of the present application.
  • the object fall detection method provided by this application can be applied to the application environment as shown in Figure 1.
  • the millimeter wave radar 102 communicates with the terminal 104 through the network.
  • the millimeter wave radar 102 is used to emit electromagnetic waves.
  • the terminal 104 first obtains the echo signal formed after the electromagnetic wave emitted by the millimeter wave radar is reflected by the object; and then performs Fourier transform processing on the echo signal. and cluster analysis to obtain the point cloud data of the target object, and perform signal strength analysis on the point cloud data based on the location information and signal strength information in the point cloud data to determine the centroid position of the point cloud cluster corresponding to the point cloud data. ; Then based on the position of the center of mass, calculate the characteristic angle of the center of mass relative to the feature point; finally, determine the fall detection result of the target object based on the change of the characteristic angle.
  • the object fall detection method provided by this application, if the computing processing capability of the millimeter wave radar meets the requirements, the application environment does not need to involve the terminal 104, but only includes the millimeter wave radar 102, which is emitted by the millimeter wave radar 102. Electromagnetic waves are detected, and fall detection is performed based on the echo signal formed after the electromagnetic waves are reflected by the object, and the fall detection result of the target object is obtained.
  • the terminal 104 includes, but is not limited to, a desktop computer, a laptop computer, a smartphone, a tablet computer, an Internet of Things device, and a portable wearable device.
  • the IoT device can be a smart speaker, smart TV, smart air conditioner, smart car device, etc.; the portable wearable device can be a smart watch, smart bracelet, head-mounted device, etc.
  • the millimeter wave radar 102 and the terminal 104 can be connected directly or indirectly through wired or wireless communication methods, which is not limited in this application.
  • an object fall detection method is provided.
  • the method is applied to the millimeter wave radar 102 for illustration. It can be understood that the method can also be applied to the terminal 104. , can also be applied to a system including the millimeter wave radar 102 and the terminal 104, and is implemented through the interaction between the millimeter wave radar 102 and the terminal 104.
  • the method includes the following steps:
  • Step S201 Obtain the echo signal formed after the electromagnetic wave emitted by the millimeter wave radar is reflected by the object.
  • millimeter wave radar refers to an electronic device that can detect target objects by emitting electromagnetic waves in the millimeter wave band.
  • Objects refer to substances with a certain shape that exist in the monitoring area and can reflect electromagnetic waves. Taking the case where the monitoring area is indoors as an example, the objects can be walls, furniture, electrical appliances, lamps, people, etc.
  • millimeter wave radar emits electromagnetic waves to the monitoring area through a transmitting antenna, and the electromagnetic waves are reflected by objects in the monitoring area to form echo signals.
  • the millimeter wave radar then receives the echo signal through the receiving antenna.
  • the transmitting antenna and the receiving antenna can also be integrated into a transmitting and receiving antenna to realize the functions of transmitting electromagnetic wave signals in the millimeter wave band and receiving echo signals.
  • the installation position and electromagnetic wave emission angle of the millimeter wave radar can also be set so that the electromagnetic waves can be radiated to the entire monitoring area as much as possible.
  • a millimeter-wave radar can be installed in a corner of a wall and emit electromagnetic waves diagonally downward or upward; another example can be based on the height of the target object, and the millimeter-wave radar can be installed on a wall whose height is smaller than the height of the target object and emit it horizontally outward. electromagnetic waves.
  • the installation height of the millimeter wave radar is half the height of the target object to ensure that the target object can be effectively detected.
  • Step S203 Perform Fourier transform processing and cluster analysis on the echo signal to obtain point cloud data of the target object.
  • Fourier transform is a process of expressing functions that meet certain conditions into linear combinations of trigonometric functions or their integrals.
  • the Fourier transform may be a continuous Fourier transform or a discrete Fourier transform.
  • Clustering is the process of dividing a collection of physical or abstract objects into classes consisting of similar objects.
  • the specific algorithm of this cluster analysis can be K nearest neighbor clustering, k-means clustering, density clustering and other algorithms.
  • the point cloud data of the target object refers to data composed of associated information representing the point cloud of the target object.
  • the point cloud data may include location information, signal strength information, speed information, angle information, etc. associated with each point cloud.
  • the echo signal of the target object can be separated from the echo signal, and information such as the position, signal strength, and speed of the target object can be obtained.
  • the position information can be characterized by three-dimensional spatial coordinates, and the signal strength information can be characterized by signal-to-noise ratio.
  • Step S205 According to the position information and signal strength information in the point cloud data, perform signal strength analysis on the point cloud data to determine the centroid position of the point cloud cluster corresponding to the point cloud data.
  • point cloud data includes location information and signal strength information associated with each point cloud.
  • the point cloud cluster corresponding to the point cloud data is the point cloud cluster composed of each point cloud in the point cloud data.
  • the centroid of a point cloud cluster refers to an imaginary point used to represent the position of the point cloud cluster.
  • the millimeter wave radar performs signal strength analysis on the point cloud data based on the signal strength information in the point cloud data to obtain the signal strength analysis result, and then combines the signal strength analysis result with the position information in the point cloud data to determine the point.
  • the centroid position of the point cloud cluster corresponding to the cloud data is the point cloud cluster composed of each point cloud in the point cloud data.
  • the centroid of a point cloud cluster refers to an imaginary point used to represent the position of the point cloud cluster.
  • the millimeter wave radar performs signal strength analysis on the point cloud data based on the signal strength information in the point cloud data to obtain the signal strength analysis result, and then combines the signal strength analysis result with the position information in the point cloud data to determine the
  • the specific way to determine the centroid position of the point cloud cluster corresponding to the point cloud data is not unique by combining the signal strength analysis results and the position information in the point cloud data.
  • point clouds with weak signal strength can be filtered out, and the average or median of the position information of point clouds whose signal strength meets the strength threshold condition can be determined as the centroid position of the point cloud cluster; another example can be based on each point cloud.
  • different point clouds are assigned corresponding weights, and then based on the weights, the positions of each point cloud are weighted and summed to obtain the centroid position of the point cloud cluster.
  • Step S207 Based on the position of the center of mass, calculate the characteristic angle of the center of mass relative to the feature point.
  • feature points refer to reference points with relatively fixed positions.
  • the feature point can be a fixed position in the monitoring space, such as a corner position, or a position determined based on millimeter wave radar, such as the installation position of millimeter wave radar.
  • millimeter wave radar can calculate the characteristic angle of the center of mass relative to the feature point based on the position of the center of mass and the position of the feature point.
  • the specific type of the characteristic angle is not unique. It can refer to the angle between the line connecting the characteristic point and the center of mass and any certain straight line.
  • the determined straight line may be a coordinate axis in a three-dimensional coordinate system.
  • the characteristic angle of the center of mass A relative to the feature point O can be any one of ⁇ OAB, ⁇ OAC, and ⁇ OAD.
  • the feature point is the origin of radar coordinates; the calculation formula of the feature angle is:
  • ⁇ is the characteristic angle
  • x, y and z jointly represent the position of the center of mass. That is, the position of the center of mass is (x, y, z), the feature point O in Figures 3 and 4 is the origin of the radar coordinates, and the feature angle ⁇ is ⁇ OAB.
  • the feature points are defined as The radar coordinate origin, because the position information determined based on the echo signal of the millimeter wave radar corresponds to the coordinate system of the radar coordinate origin. Therefore, in the process of calculating the characteristic angle, there is no need to perform coordinate conversion, which can simplify the calculation process of the characteristic angle. It is helpful to improve data processing efficiency.
  • Step S209 Determine the fall detection result of the target object based on changes in the characteristic angle.
  • the fall detection results may include fall occurrence and non-fall occurrence.
  • the fall determination conditions used to characterize the correspondence between the change of the characteristic angle and the fall detection result can be determined.
  • the fall determination condition may be that the change trend of the current feature angle relative to the historical feature angle satisfies the set trend determination condition, and/or the change amount of the current feature angle relative to the historical feature angle satisfies the set change amount determination condition.
  • the setting trend judgment condition corresponding to Figure 3 may be that the characteristic angle becomes larger
  • the setting trend judgment condition corresponding to Figure 4 may be that the characteristic angle becomes smaller.
  • the change amount of the characteristic angle may gradually increase.
  • the set change amount determination condition corresponding to Figure 3 can be the current
  • the first change amount of the characteristic angle relative to the first historical characteristic angle calculated last time is greater than the second change amount of the first historical characteristic angle relative to the second historical characteristic angle calculated last time.
  • the change of the currently calculated characteristic angle relative to the historical characteristic angle can be compared with the fall determination condition.
  • the change of the characteristic angle meets the fall determination condition, we obtain Fall detection results when the target object falls.
  • the millimeter wave radar can also output the fall detection result in at least one of various forms such as pictures, text, and voice, or a combination of at least two of the above forms, and the fall detection result
  • the output object can be a storage device, a display device or a communication device.
  • millimeter wave radar can also output fall detection results to the terminal through the communication device. Millimeter wave radar can also output warning information when the fall detection result is a fall.
  • the above object fall detection method first obtains the echo signal formed after the electromagnetic wave emitted by the millimeter wave radar is reflected by the object, and then performs Fourier transform processing and cluster analysis on the echo signal to obtain the point cloud data of the target object, and According to the position information and signal strength information in the point cloud data, perform signal strength analysis on the point cloud data to determine the centroid position of the point cloud cluster corresponding to the point cloud data, and then calculate the centroid relative to the feature point based on the centroid position. characteristic angle, and finally determine the fall detection result of the target object based on the change of the characteristic angle.
  • the point cloud data of the target object can be identified, and the signal strength analysis of the point cloud data can be performed to determine the centroid position of the point cloud cluster corresponding to the point cloud data.
  • the interference of other objects other than the target object in the environment is eliminated, and the entire fall detection process does not require the target object to wear any auxiliary sensing device. It has strong environmental adaptability and is conducive to improving the accuracy of the target object's fall detection results.
  • fall detection based on millimeter wave radar has no risk of privacy leakage, does not require wearing, and has strong environmental adaptability, which is conducive to expanding the application scenarios of fall detection methods.
  • step S203 includes: combining at least one of attribute characteristics or motion characteristics of the target object in the object, performing Fourier transform processing and cluster analysis on the echo signal to obtain point cloud data of the target object. .
  • the objects that reflect electromagnetic waves to form echo signals include the target object to be detected for fall.
  • the process of obtaining the point cloud data of the target object based on the echo signal formed by the reflection of the object is actually screening to obtain the point cloud of the target object. data process.
  • the corresponding clustering algorithm can be selected according to the attribute characteristics of the target object to implement point cloud data screening. .
  • the velocity information and position information of the corresponding point cloud data will change differently. Therefore, point cloud data can also be filtered based on the motion characteristics of the target object. For example, furniture, walls, etc. usually do not move, and electrical appliances such as electric fans will move strictly according to the preset cycle. Based on this, millimeter wave radar can combine at least one of the attribute characteristics or motion characteristics of the target object in the object, perform Fourier transform processing and cluster analysis on the echo signal, and filter out the point cloud data of the target object.
  • millimeter wave radar first clusters point cloud data into multiple point cloud clusters based on density clustering algorithm, then determines the motion characteristics of each point cloud cluster, and matches the motion characteristics with the motion characteristics of the target object to obtain a point cloud.
  • Clusters are determined as point cloud clusters containing point cloud data of the target object.
  • point cloud clusters of multiple target objects in the monitoring space can be clustered.
  • the motion characteristics of each point cloud cluster can be determined, and then by matching the motion characteristics with the motion characteristics of the target object, the point cloud cluster of the target object can be determined.
  • the point cloud data contained in this point cloud cluster is the point cloud data of the target object.
  • At least one of the attribute characteristics or motion characteristics of the target object is comprehensively considered to obtain the point cloud data of the target object, which is beneficial to improving the accuracy of the point cloud data screening results, thereby improving the accuracy of the fall detection results. sex.
  • Fourier transform processing and cluster analysis are performed on the echo signal in combination with at least one of attribute characteristics or motion characteristics of the target object in the object to obtain point cloud data of the target object, including:
  • the echo signal is processed by fast Fourier transform to obtain the position information of the object; according to the signal-to-noise ratio of the echo signal, the signal strength information of each position is determined; combined with at least one of the attribute characteristics or motion characteristics of the target object in the object First, perform cluster analysis on point cloud data containing location information and signal strength information to obtain point cloud data of the target object.
  • millimeter wave radar performs distance-dimensional fast Fourier transform on the echo signal to obtain the relative distance between the object and the radar transmitting antenna, thereby determining the position information of the object, and based on the echo signal corresponding to each position.
  • Signal-to-noise ratio determines the signal strength information at each location. It can be understood that the greater the signal-to-noise ratio, the greater the energy of the echo signal at the corresponding location and the greater the signal strength.
  • millimeter-wave radar can also perform fast Fourier transform of the velocity dimension on the position information to obtain the velocity information of the object, and then synthesize the position information and velocity information to generate a range Doppler detection matrix, and calculate the range
  • the Puller detection matrix performs constant false alarm detection to obtain multiple target data, and then performs fast Fourier transform on the target data to determine the angle information.
  • the point cloud data can include information such as position, signal strength, speed, and angle.
  • the millimeter wave radar combines at least one of the attribute characteristics or motion characteristics of the target object in the object to perform cluster analysis on the point cloud data, so that the point cloud data of the target object can be obtained by screening the point cloud data.
  • the point cloud data of the target object is obtained by performing fast Fourier transform processing and cluster analysis on the echo signal, so as to reduce the amount of calculation and improve the data processing efficiency.
  • step S205 includes: performing signal strength analysis on the point cloud data according to the position information and signal strength information in the point cloud data, and obtaining the respective position and signal strength of each point cloud in the point cloud data; based on each point The position and signal strength of each cloud determine the centroid position of the point cloud cluster composed of each point cloud.
  • the signal strength is characterized by the signal-to-noise ratio.
  • Point cloud data corresponds to multiple point clouds, and each point cloud corresponds to unique position information and signal strength information.
  • the point cloud cluster corresponding to the point cloud data is the point cloud composed of each point cloud in the point cloud data. cluster.
  • the process of analyzing the signal strength of point cloud data to determine the centroid position is essentially to determine the centroid position of the point cloud cluster composed of each point cloud based on the corresponding position and signal strength of each point cloud in the point cloud data.
  • millimeter wave radar can perform signal strength analysis on the point cloud data based on the position information and signal strength information in the point cloud data, and obtain the corresponding position and signal strength of each point cloud in the point cloud data.
  • centroid position of the point cloud cluster composed of each point cloud is determined.
  • the point cloud whose signal strength meets the intensity threshold condition can be determined as the selected point cloud, and the average or median of the respective positions of each selected point cloud can be determined as the centroid position of the point cloud cluster.
  • the position of the center of mass is determined by combining the position information and the signal strength information, which can reduce the impact of noise information and help improve the accuracy of the center of mass position, thereby improving the accuracy of fall detection results.
  • determining the centroid position of the point cloud cluster composed of each point cloud based on the respective position and signal strength of each point cloud includes: determining the respective centroid position of each point cloud based on the corresponding signal strength information of each point cloud.
  • Weight According to the weight of each point cloud, perform a weighted summation of the corresponding position information of each point cloud to determine the centroid position of the point cloud cluster formed by the point cloud.
  • each point cloud corresponds to one point cloud data
  • each point cloud data contains location information and signal strength information.
  • millimeter wave radar can determine the weight of each point cloud based on the signal strength information corresponding to each point cloud, and then perform a weighted sum of the position information corresponding to each point cloud data based on the weight of each point cloud.
  • the centroid position of the point cloud cluster composed of point cloud data can be determined.
  • the specific way to determine the weight of each point cloud is not unique. For example, when the signal strength of a certain point cloud does not meet the intensity threshold condition, the weight of the point cloud can be determined to be zero, and the weights of the remaining point clouds that meet the intensity threshold condition can be set to the same value; for another example, you can The corresponding relationship between the signal strength interval and the weight is established. For each point cloud, the weight of the point cloud is determined based on the signal strength interval where the signal strength of the point cloud is located.
  • centroid of the point cloud cluster is determined based on the signal strength of each point cloud.
  • the algorithm is simple and is conducive to improving data processing efficiency.
  • determining the weight of each point cloud based on the signal strength information corresponding to each point cloud includes: determining the total energy of the point cloud cluster composed of each point cloud based on the signal strength of each point cloud; For each point cloud, the ratio of the signal strength of the point cloud to the total energy is determined as the weight of the point cloud.
  • the total energy of the point cloud cluster composed of each point cloud is the sum of the signal strengths of each point cloud.
  • the millimeter wave radar superimposes the signal intensity of each point cloud to obtain the total energy of the point cloud cluster composed of each point cloud. Then, for each point cloud, the ratio of the signal intensity of the point cloud to the total energy is, Determine the weight of this point cloud.
  • the weight of the point cloud is determined based on the energy distribution of each point cloud, which can automatically filter out point clouds with smaller energy, which is beneficial to improving the accuracy of the position information of the center of mass.
  • step S209 includes: determining that the fall detection result of the target object is that a fall occurs when the change of the characteristic angle within a set number of consecutive time frames satisfies the fall determination condition.
  • the fall determination condition is that the proportion of positive or negative values in the difference between each feature angle and the historical feature angle of the previous frame is greater than the set proportion, or the change in each difference relative to the previous difference is increased.
  • the time frame refers to the time information corresponding to the feature angle. Since millimeter wave radar emits electromagnetic waves according to a certain period during fall detection, and the characteristic angle is calculated based on the echo signal of a certain period of electromagnetic waves, therefore, the characteristic angle corresponds to a unique time frame. This time frame corresponds to the emission time of the electromagnetic wave.
  • the millimeter wave radar determines that the fall detection result of the target object is a fall when the change of the characteristic angle within a set number of continuous time frames satisfies the fall determination condition; within a set number of continuous time frames, the characteristic angle If the change does not meet the fall determination conditions, the fall detection result of the target object is determined to be that no fall has occurred.
  • the characteristic point is the radar coordinate origin O and the characteristic angle is the angle OAB.
  • the echo signal formed after the electromagnetic wave emitted by the radar is reflected by the target object will also change, reflected in the point
  • the cloud data shows: Even if attitude changes are not taken into account, the signal intensity distribution of each point cloud will also change. Then, the centroid position determined based on the respective signal strengths of each point cloud will also change. That is, during the translation process of the target object, the centroid position of the target object will not remain on the plane CAD.
  • the impact of translation on the feature angle will be volatile and irregular.
  • the fall detection result of the target object is determined to be a fall, and the fall determination condition is the historical feature angle between each characteristic angle and the previous frame.
  • the proportion of positive or negative values in the differences is greater than the set proportion, or the change of each difference relative to the previous difference increases, which can reflect the continuous and regular change process of the characteristic angle during the fall. It can avoid misjudgment of falls caused by translation of the target object.
  • determining that the fall detection result of the target object is that a fall has occurred includes: obtaining a set number of consecutive time frames. The calculated feature angle; based on the corresponding time frame of each feature angle, when the feature angle has the historical feature angle of the previous frame, calculate the difference between the feature angle and the historical feature angle of the previous frame; in each difference When the fall determination conditions are met, the detection result that the target object has fallen is obtained.
  • millimeter-wave radar obtains characteristic angles calculated within a set number of consecutive time frames.
  • the specific value of the set number is not unique, for example, it can be 10, 12, 14, 15 or 18, etc.
  • the millimeter wave radar subtracts the historical characteristic angle of the previous frame from the characteristic angle when the characteristic angle has the historical characteristic angle of the previous frame, and calculates the difference between the characteristic angle and the previous frame. Difference of historical feature angles of frames. It can be understood that when the number is set to 12, 11 differences will be obtained.
  • the millimeter-wave radar compares each difference value with the fall determination condition. If each difference value meets the fall determination condition, the detection result that the target object has fallen is obtained.
  • the fall determination condition can be that the proportion of positive values in each difference is greater than the set proportion. For example, when the set number is 12, 11 are obtained The differences are all positive; it can also be that the change in the difference between each difference relative to the previous difference increases. For example, when the set quantity is 12, 11 differences will be obtained and then the difference can be calculated. 10 difference changes were obtained, and each difference change was positive.
  • the fall determination is performed based on the characteristic angles calculated within a set number of consecutive time frames, which is beneficial to improving the accuracy of the fall detection results.
  • the object fall detection method includes:
  • Step S501 obtain the echo signal formed after the electromagnetic wave emitted by the millimeter wave radar is reflected by the object;
  • Step S502 perform fast Fourier transform processing on the echo signal to obtain the position information of the object
  • Step S503 determine the signal strength information of each location according to the signal-to-noise ratio of the echo signal
  • Step S504 combine at least one of the attribute characteristics or motion characteristics of the target object in the object, perform cluster analysis on the point cloud data containing position information and signal strength information, and obtain point cloud data of the target object;
  • Step S505 perform signal strength analysis on the point cloud data according to the position information and signal strength information in the point cloud data, and obtain the respective position and signal strength of each point cloud in the point cloud data;
  • Step S506 Based on the signal strength of each point cloud, determine the total energy of the point cloud cluster composed of each point cloud;
  • Step S507 for each point cloud, determine the weight of the point cloud by determining the ratio of the signal strength of the point cloud to the total energy;
  • Step S508 perform a weighted sum of the respective positions of each point cloud according to the respective weights of each point cloud, and determine the centroid position of the point cloud cluster composed of each point cloud;
  • Step S509 based on the position of the center of mass, calculate the characteristic angle of the center of mass relative to the feature point;
  • Step S510 obtain the characteristic angle calculated within a set number of consecutive time frames
  • Step S511 based on the time frame corresponding to each feature angle, if the feature angle has a historical feature angle of the previous frame, calculate the difference between the feature angle and the historical feature angle of the previous frame;
  • Step S512 When each difference value meets the fall determination condition, a detection result that the target object falls is obtained.
  • the millimeter wave radar is installed on a wall 1.5 meters above the ground, and the normal direction of the radar is parallel to the ground.
  • millimeter-wave radar transmits millimeter-wave band radio frequency signals to the monitoring area through a multiple-in multiple-out (MIMO) radio frequency transceiver antenna, and at the same time receives the radio frequency signal scattered by obstacles in the monitoring area. echo signal.
  • the echo signal is mixed with the transmitted signal to form an intermediate frequency signal, and the intermediate frequency signal is sampled by an ADC (Analog to Digital Converter, analog-to-digital conversion) to obtain sampling information.
  • ADC Analog to Digital Converter, analog-to-digital conversion
  • Doppler processing (2D-FFT) is performed on the position information to obtain the speed information of the target object.
  • the position information and velocity information on each channel are synthesized to generate a range Doppler detection matrix, and then the detection matrix is subjected to Constant False-Alarm Rate (CFAR) processing and angle calculation to filter out false targets.
  • CFAR Constant False-Alarm Rate
  • the point cloud data set includes position information, angle information, Doppler velocity information, and signal strength information.
  • the position information is represented by the coordinate values of the radar three-dimensional coordinate system
  • the signal strength information is represented by the signal-to-noise ratio.
  • P i [X i ,Y i ,Z i ,V i ,S i ] represents a point cloud data
  • (X i ,Y i ,Z i ) is the three-dimensional coordinates of point i
  • V i is the coordinate of point i
  • Speed S i is the signal-to-noise ratio of point i.
  • the selected human body point cloud data will form a point cloud cluster. Based on the energy distribution of each point cloud data in the point cloud cluster, the centroid position of the point cloud cluster can be calculated.
  • the pitch angle corresponding to a time frame can be obtained.
  • the w1 array is filled.
  • calculate the difference ⁇ between adjacent elements in the w1 window and save the difference in the form w2.
  • embodiments of the present application also provide an object fall detection device for implementing the above-mentioned object fall detection method.
  • the solution to the problem provided by this device is similar to the solution recorded in the above method. Therefore, the specific limitations in the embodiments of one or more object fall detection devices provided below can be found in the above description of the object fall detection method. Limitations will not be repeated here.
  • an object fall detection device 800 including: an echo signal acquisition module 801, a target object echo signal analysis module 802, a centroid position determination module 803, and a characteristic angle calculation module. 804 and fall detection result determination module 805, wherein:
  • the echo signal acquisition module 801 is used to acquire the echo signal formed after the electromagnetic wave emitted by the millimeter wave radar is reflected by the object;
  • the echo signal analysis module 802 is used to perform Fourier transform processing and cluster analysis on the echo signal to obtain point cloud data of the target object;
  • the centroid position determination module 803 is used to perform signal strength analysis on the point cloud data based on the position information and signal strength information in the point cloud data, and determine the centroid position of the point cloud cluster corresponding to the point cloud data;
  • Feature angle calculation module 804 is used to calculate the feature angle of the centroid relative to the feature point based on the centroid position
  • the fall detection result determination module 805 is used to determine the fall detection result of the target object according to the change of the characteristic angle.
  • centroid position determination module 803 is specifically used to:
  • the position information and signal strength information in the point cloud data perform signal strength analysis on the point cloud data to obtain the respective position and signal strength of each point cloud in the point cloud data; based on the respective position and signal strength of each point cloud, determine each point cloud.
  • the centroid position of the point cloud cluster formed by the point cloud; the signal strength is characterized by the signal-to-noise ratio.
  • the centroid position determination module 803 includes:
  • a weight determination unit used to determine the respective weights of each point cloud based on the respective signal strengths of each point cloud
  • the position information determination unit is used to perform a weighted sum of the respective positions of each point cloud based on the respective weights of each point cloud, and determine the centroid position of the point cloud cluster composed of each point cloud.
  • the weight determination unit is specifically used to:
  • the total energy of the point cloud cluster composed of each point cloud is determined; for each point cloud, the ratio of the signal intensity of the point cloud to the total energy is determined as the weight of the point cloud.
  • the echo signal analysis module 802 is specifically used to:
  • Fourier transform processing and cluster analysis are performed on the echo signal to obtain point cloud data of the target object.
  • the echo signal analysis module 802 is specifically used to:
  • One method is to perform cluster analysis on point cloud data containing location information and signal strength information to obtain point cloud data of the target object.
  • the feature point is the origin of radar coordinates; the calculation formula of the feature angle is:
  • ⁇ is the characteristic angle
  • x, y and z jointly represent the position of the center of mass.
  • the fall detection result determination module 805 is specifically used to:
  • the fall detection result of the target object is determined to be a fall;
  • the fall determination condition is the difference between each feature angle and the historical feature angle of the previous frame.
  • the proportion of positive or negative values in the values is greater than the set proportion, or the change in the difference between each difference value relative to the previous difference value increases.
  • the fall detection result determination module 805 is specifically used to:
  • Each module in the above object fall detection device can be implemented in whole or in part by software, hardware and combinations thereof.
  • Each of the above modules can be embedded in or independent of the processor of the millimeter wave radar in the form of hardware, or can be stored in the memory of the millimeter wave radar in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.
  • a millimeter wave radar is provided.
  • the millimeter wave radar includes a radar signal transceiver device 901 and a processor 902 .
  • the radar signal transceiving device 901 is used to emit electromagnetic waves and receive echo signals formed after the electromagnetic waves are reflected by objects; the processor 902 is used to implement the steps in the above object fall detection method.
  • the radar signal transceiving device 901 may include a transmitting device and a receiving device, or may be an integrated device that integrates transmitting and receiving. Furthermore, the radar signal transceiver device 901 may also be a radio frequency transceiver device.
  • the processor 902 may be a hardware module including various processing chips and peripheral circuits, and having logical operation functions.
  • the processing chip can be a microcontroller, a DSP (Digital Signal Process, digital signal processing) chip or an FPGA (Field Programmable Gate Array, field programmable logic gate array) chip.
  • the radar signal transceiving device 901 is used to emit electromagnetic waves and receive echo signals formed after the electromagnetic waves are reflected by objects; the processor 902 is used to implement the steps in the above object fall detection method.
  • the millimeter wave radar may also include a communication device 903 and a power supply device 904 .
  • the communication device 903 may be a wired communication device or a wireless communication device.
  • the wired communication device may be a bus communication device, such as a 485 communication device, a CAN communication device or an RS232 communication device.
  • the wireless communication device may be a Bluetooth communication device, a wireless communication device or a cellular communication device.
  • the power supply device 904 may be a power supply plug, used to obtain electric energy from an external power source, or may be a device containing an energy storage device that can output electric energy to the outside.
  • the energy storage device may be an energy storage battery pack or a supercapacitor.
  • the processor 902 can be connected to the radar signal transceiving device 901 and the communication device 903 respectively through the serial port.
  • millimeter wave radar can also include storage devices, display devices, warning devices, etc.
  • the storage device is used to store a computer program.
  • the processor 902 processes the computer program, the above-mentioned object fall detection method is implemented.
  • the display device is used to display a fall detection result, and the warning device is used to output warning information when the fall detection result is a fall.
  • the structure shown in Figure 9 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the millimeter wave radar to which the solution of the present application is applied.
  • the millimeter wave radar Radars may include more or fewer components than shown, or certain components may be combined, or may have a different arrangement of components.
  • a computer-readable storage medium on which a computer program is stored.
  • the steps in the above object fall detection method are implemented.
  • a computer program product including a computer program that implements the steps in the above object fall detection method when executed by a processor.
  • the computer program can be stored in a non-volatile computer-readable storage.
  • the computer program when executed, may include the processes of the above method embodiments.
  • Any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory.
  • Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetic variable memory (Magnetoresistive Random Access Memory (MRAM), ferroelectric memory (Ferroelectric Random Access Memory, FRAM), phase change memory (Phase Change Memory, PCM), graphene memory, etc.
  • Volatile memory may include random access memory (Random Access Memory, RAM) or external cache memory, etc.
  • RAM Random Access Memory
  • RAM random access memory
  • RAM Random Access Memory
  • the databases involved in the various embodiments provided in this application may include at least one of a relational database and a non-relational database.
  • Non-relational databases may include blockchain-based distributed databases, etc., but are not limited thereto.
  • the processors involved in the various embodiments provided in this application may be general-purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to this.

Landscapes

  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Physiology (AREA)
  • Veterinary Medicine (AREA)
  • Public Health (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • General Health & Medical Sciences (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Medical Informatics (AREA)
  • General Physics & Mathematics (AREA)
  • Surgery (AREA)
  • Animal Behavior & Ethology (AREA)
  • Electromagnetism (AREA)
  • Oral & Maxillofacial Surgery (AREA)
  • Dentistry (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychiatry (AREA)
  • Signal Processing (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

本申请涉及一种对象跌倒检测方法、装置、毫米波雷达、计算机可读存储介质和计算机程序产品。该方法包括：获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置；基于质心位置，计算质心相对于特征点的特征角；根据特征角的变化情况确定目标对象的跌倒检测结果。采用上述方法可以提高跌倒检测结果的准确性。

Description

对象跌倒检测方法、装置、毫米波雷达和存储介质
相关申请的交叉引用
本申请引用于2022年6月10日递交的名称为“对象跌倒检测方法、装置、毫米波雷达和存储介质”的第202210651286.4号中国专利申请，其通过引用被全部并入本申请。
技术领域
本申请涉及跌倒检测技术领域，特别是涉及一种对象跌倒检测方法、装置、毫米波雷达、计算机可读存储介质和计算机程序产品。
背景技术
众所周知，随着年龄的增加，人体机能水平逐渐衰退，健康风险随之增加，例如可能会发生跌倒等意外伤害。一旦发生跌倒，若不能及时被人发现并采取相应的救护措施，可能会引起神经损伤和瘫痪等严重的身体伤害。基于此，有必要对目标对象，特别是独居老人进行跌倒检测，及时发现目标对象的跌倒行为。
传统的对象跌倒检测方法，由目标对象佩戴可穿戴传感装置，通过装置中的传感器感知人体在三维空间的多个方向的速度或者加速度，进行跌倒判断。然而，采用上述方法，一方面，由于可穿戴传感装置的佩戴舒适性差等原因，目标对象在洗漱等易发生跌倒的时间，可能会存在未佩戴可穿戴传感装置的情况，无法进行有效的跌倒检测；另一方面，受跌倒判断算法的制约，甩手等存在跌倒特征的动作容易被误判为跌倒行为。因此，传统的对象跌倒检测方法，存在检测结果准确性差的缺点。
申请内容
根据本申请公开的各种实施例，提供一种对象跌倒检测方法、装置、毫米波雷达、计算机可读存储介质和计算机程序产品，以提高对象跌倒检测结果的准确性。
一种对象跌倒检测方法，包括：
获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定该点云数据所对应的点云簇的质心位置；基于质心位置，计算质心相对于特征点的特征角；及根据特征角的变化情况确定目标对象的跌倒检测结果。
上述对象跌倒检测方法，通过对回波信号进行傅里叶变换处理和聚类分析可以识别出目标对象的点云数据，对该点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置，可以排除环境中目标对象以外其他物体的干扰，并且整个跌倒检测过程无需目标对象佩戴任何辅助传感装置，环境适应性强，有利于提高目标对象的跌倒检测结果的准确性。
在一些实施例中，根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定该点云数据所对应的点云簇的质心位置，包括：
根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，获得点云数据中每一点云各自的位置和信号强度；及基于各点云各自的位置和信号强度，确定各点云所构成的点云簇的质心位置；信号强度通过信噪比表征。
上述实施例中，结合位置信息和信号强度信息确定质心位置，可以降低噪声信息的影响，有利于提高质心位置的准确性，进而提升跌倒检测结果的准确性。
在一些实施例中，根据点云数据中的位置信息和信号强度信息，基于各点云各自的位 置和信号强度，确定各点云所构成的点云簇的质心位置，包括：
基于各点云各自的信号强度，确定各点云各自的权重；及根据各点云各自的权重，对各点云各自的位置进行加权求和，确定各点云所构成的点云簇的质心位置。
上述实施例中，基于各点云的信号强度确定点云簇的质心，算法简单，有利于提高数据处理效率。
在一些实施例中，基于各点云各自对应的信号强度信息，确定各点云各自的权重，包括：
基于各点云各自的信号强度，确定各点云所构成的点云簇的总能量；及针对每一点云，将该点云的信号强度与总能量的比值，确定为该点云的权重。
上述实施例中，基于各点云数据的能量分布确定点云数据的权重，可以自动过滤掉能量较小的点云，有利于提高质心的位置信息的准确性。
在一些实施例中，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据，包括：
结合物体中的目标对象的属性特征或运动特征中的至少一种，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据。
上述实施例中，综合考虑的目标对象的属性特征或运动特征中的至少一种，得到目标对象的点云数据，有利于提高点云数据与目标对象的匹配度，进而提升跌倒检测结果的准确性。
在一些实施例中，结合物体中的目标对象的属性特征或运动特征中的至少一种，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据，包括：
对回波信号进行快速傅里叶变换处理，获得物体的位置信息；根据回波信号的信噪比，确定各位置的信号强度信息；及结合物体中的目标对象的属性特征或运动特征中的至少一种，对包含位置信息和信号强度信息的点云数据进行聚类分析，得到目标对象的点云数据。
上述实施例中，通过对回波信号进行快速傅里叶变换处理和聚类分析得到目标对象的点云数据，以减少计算量、提高数据处理效率。
在一些实施例中，特征点为雷达坐标原点；特征角的计算公式为：
式中，θ为特征角，x、y和z共同表征质心位置。
上述实施例中，将特征点定义为雷达坐标原点，由于基于毫米波雷达的回波信号确定的位置信息，与雷达坐标原点的坐标系对应，因此，在计算特征角的过程中，无需进行坐标转换，可以简化特征角的计算过程，有利于提高数据处理效率。
在一些实施例中，根据特征角的变化情况确定目标对象的跌倒检测结果，包括：
在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒；跌倒判定条件为各特征角与上一帧的历史特征角的差值中正值或负值的比例大于设定比例，或者，各差值相对于前一差值的差值变化量增大。
上述实施例中，基于设定数量的连续时间帧内计算得到的特征角进行跌倒判断，有利于提高跌倒检测结果的准确性。
在一些实施例中，在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒，包括：
获取设定数量的连续时间帧内计算得到的特征角；基于各特征角各自对应的时间帧，在特征角存在上一帧的历史特征角的情况下，计算特征角与上一帧的历史特征角的差值；及在各差值满足跌倒判定条件的情况下，得到目标对象发生跌倒的检测结果。
上述实施例中，在连续时间帧内的各差值满足跌倒判定条件的情况下确定目标对象发生跌倒，有利于提高跌倒检测结果的准确性。
一种对象跌倒检测装置，包括：
回波信号获取模块，用于获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；
目标对象回波信号分析模块，用于对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；
质心位置确定模块，用于根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定该点云数据所对应的点云簇的质心位置；
特征角计算模块，用于基于质心位置，计算质心相对于特征点的特征角；及
跌倒检测结果确定模块，用于根据特征角的变化情况确定目标对象的跌倒检测结果。
一种计算机设备，包括存储器及一个或多个处理器，存储器中储存有计算机可读指令，计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：
获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定该点云数据所对应的点云簇的质心位置；基于质心位置，计算质心相对于特征点的特征角；及根据特征角的变化情况确定目标对象的跌倒检测结果。
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：
获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定该点云数据所对应的点云簇的质心位置；基于质心位置，计算质心相对于特征点的特征角；及根据特征角的变化情况确定目标对象的跌倒检测结果。
一种计算机程序产品，包括计算机程序，该计算机程序被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：
获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定该点云数据所对应的点云簇的质心位置；基于质心位置，计算质心相对于特征点的特征角；及根据特征角的变化情况确定目标对象的跌倒检测结果。
上述跌倒检测装置、计算机设备、计算机可读存储介质和计算机程序产品，通过对回波信号进行傅里叶变换处理和聚类分析可以识别出目标对象的点云数据，对该点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置，可以排除环境中目标对象以外其他物体的干扰，并且整个跌倒检测过程无需目标对象佩戴任何辅助传感装置，环境适应性强，有利于提高目标对象的跌倒检测结果的准确性。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的附图。
图1为本申请一些实施例中对象跌倒检测方法的应用环境图；
图2为本申请一些实施例中对象跌倒检测方法的流程示意图；
图3为本申请一些实施例中特征点与质心的相对位置示意图；
图4为本申请另一些实施例中特征点与质心的相对位置示意图；
图5为本申请另一些实施例中对象跌倒检测方法的流程示意图；
图6为本申请一些实施例中对象跌倒检测过程的示意图；
图7为本申请一些实施例中跌倒判断算法的流程示意图；
图8为本申请一些实施例中对象跌倒检测装置的结构框图；
图9为本申请一些实施例中毫米波雷达的内部结构图。
具体实施方式
为了使本申请的技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
除非另有定义，本文所使用的所有的技术和科学术语与属于本申请的技术领域的技术人员通常理解的含义相同；本文中所使用的术语只是为了描述具体的实施例的目的，不是旨在于限制本申请；本申请的说明书和权利要求书及上述附图说明中的术语“包括”和“具有”以及它们的任何变形，意图在于覆盖不排他的包含。
在本申请实施例的描述中，技术术语“第一”“第二”等仅用于区别不同对象，而不能理解为指示或暗示相对重要性或者隐含指明所指示的技术特征的数量、特定顺序或主次关系。在本申请实施例的描述中，“多个”的含义是两个以上，除非另有明确具体的限定。同时，在本说明书中使用的术语“和/或”包括相关所列项目的任何及所有组合。
在一个实施例中，本申请提供的对象跌倒检测方法，可以应用于如图1所示的应用环境中。其中，毫米波雷达102通过网络与终端104进行通信。毫米波雷达102用于发射电磁波，终端104在进行对象跌倒检测的过程中：先获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；再对该回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据，并根据点云数据中的位置信息和信号强度信息，对该点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置；然后基于该质心位置，计算该质心相对于特征点的特征角；最后根据该特征角的变化情况确定目标对象的跌倒检测结果。
在一个实施例中，本申请提供的对象跌倒检测方法，在毫米波雷达的计算处理能力满足要求的情况下，该应用环境不必涉及终端104，仅包括毫米波雷达102，由毫米波雷达102发射电磁波，并基于电磁波被物体反射后形成的回波信号进行跌倒检测，得到目标对象的跌倒检测结果。
其中，终端104包括但不限于是台式计算机、笔记本电脑、智能手机、平板电脑、物联网设备和便携式可穿戴设备。该物联网设备可为智能音箱、智能电视、智能空调、智能车载设备等；该便携式可穿戴设备可为智能手表、智能手环、头戴设备等。毫米波雷达102以及终端104可以通过有线或无线通信方式进行直接或间接地连接，本申请在此不做限制。
在一个实施例中，如图2所示，提供了一种对象跌倒检测方法，本实施例以该方法应用于毫米波雷达102进行举例说明，可以理解的是，该方法也可以应用于终端104，还可以应用于包括毫米波雷达102和终端104的系统，并通过毫米波雷达102和终端104的交互实现。本实施例中，该方法包括以下步骤：
步骤S201，获取毫米波雷达发射的电磁波被物体反射后形成的回波信号。
其中，毫米波雷达是指可以通过发射毫米波波段的电磁波，进行目标物体的探测的电子设备。物体是指监测区域中存在的，能够反射电磁波的，具有一定形状的物质。以监测区域为室内的情况为例，该物体可以是墙壁、家具、电器、灯具和人等等。
具体地，毫米波雷达通过发射天线向监测区域发射电磁波，该电磁波被监测区域中的物体反射后形成回波信号。毫米波雷达再通过接收天线接收该回波信号。可以理解， 该发射天线和接收天线也可以集成为收发天线，用于实现毫米波段电磁波信号的发射和回波信号的接收功能。进一步地，还可以通过对毫米波雷达的安装位置和电磁波发射角度进行设置，以使电磁波尽量辐射到整个监测区域。例如，可以将毫米波雷达安装在墙角，斜向下或斜向上发射电磁波；又如，可以根据目标对象的身高，将毫米波雷达安装于高度小于目标对象的身高的墙面，水平向外发射电磁波。在一个实施例中，毫米波雷达的安装高度为目标对象的身高的一半，以确保目标对象能被有效检测。
步骤S203，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据。
其中，傅里叶变换是将满足一定条件的函数表示成三角函数或者它们的积分的线性组合的过程。该傅里叶变换的具体可以是连续傅里叶变换或离散傅里叶变换。聚类是指将物理或抽象对象的集合分成由类似的对象组成的多个类的过程。该聚类分析的具体算法可以是K近邻聚类、k-means聚类以及密度聚类等算法。目标对象的点云数据是指由表征目标对象的点云的关联信息所构成的数据。该点云数据可以包括各点云所关联的位置信息、信号强度信息、速度信息和角度信息等等。
具体地，通过对回波信号进行傅里叶变换和聚类处理，可以从回波信号中分离出目标对象的回波信号，并获得该目标对象的位置、信号强度和速度等信息。该位置信息可以通过三维空间坐标表征，该信号强度信息可以通过信噪比表征。
步骤S205，根据点云数据中的位置信息和信号强度信息，对该点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置。
如前文所述，点云数据中包括各点云所关联的位置信息和信号强度信息。点云数据所对应的点云簇，即为点云数据中各点云所构成的点云簇。点云簇的质心是指用以表征该点云簇的位置的假想点。具体地，毫米波雷达根据点云数据中的信号强度信息，对该点云数据进行信号强度分析，得到信号强度分析结果，再结合该信号强度分析结果与点云数据中的位置信息，确定点云数据所对应的点云簇的质心位置。
需要说明的是，结合信号强度分析结果与点云数据中的位置信息，确定点云数据所对应的点云簇的质心位置的具体方式并不唯一。例如，可以过滤掉信号强度较弱的点云，将信号强度满足强度阈值条件的点云的位置信息的平均值或中位数确定为点云簇的质心位置；又如，可以基于各点云的信号强度，对不同点云分配对应的权重，再基于该权重，对各点云的位置进行加权求和，得到点云簇的质心位置。
步骤S207，基于质心位置，计算该质心相对于特征点的特征角。
其中，特征点是指位置相对固定的基准点。该特征点，可以是监测空间内的某一固定位置，如墙角位置，也可以是基于毫米波雷达确定的位置，如毫米波雷达的安装位置。具体地，与质心相同，特征点也关联有对应的特征点位置信息。由于特征点为固定点，该特征点位置通常也是固定的。基于此，毫米波雷达可以根据质心位置和特征点位置，计算该质心相对于特征点的特征角。
该特征角的具体类型并不唯一，可以是指特征点与质心的连线与任一确定直线的夹角。该确定直线，可以是三维坐标系中的坐标轴。如图3中，质心A相对于特征点O的特征角可以是∠OAB、∠OAC和∠OAD中的任意一个。
在一个实施例中，特征点为雷达坐标原点；特征角的计算公式为：
式中，θ为特征角，x、y和z共同表征质心位置。也即，质心的位置为(x,y,z)，图3和图4中的特征点O为雷达坐标原点，特征角θ为∠OAB。本实施例中，将特征点定义为 雷达坐标原点，由于基于毫米波雷达的回波信号确定的位置信息，与雷达坐标原点的坐标系对应，因此，在计算特征角的过程中，无需进行坐标转换，可以简化特征角的计算过程，有利于提高数据处理效率。
步骤S209，根据特征角的变化情况确定目标对象的跌倒检测结果。
其中，跌倒检测结果可以包括发生跌倒和未发生跌倒。基于目标对象发生跌倒时特征角的变化特点，可以确定用于表征特征角变化情况与跌倒检测结果的对应关系的跌倒判定条件。该跌倒判定条件可以是当前特征角相对于历史特征角的变化趋势满足设定趋势判定条件，和/或，当前特征角相对于历史特征角的变化量满足设定变化量判定条件。例如，特征角为∠OAB的情况下，图3对应的设定趋势判定条件可以是特征角变大，图4对应的设定趋势判定条件可以是特征角变小。又如，由于跌倒过程中，受重力加速度的影响，特征角的变化量可能会逐步增大，基于此，特征角为∠OAB的情况下，图3对应的设定变化量判定条件可以是当前特征角相对于上一次计算得到的第一历史特征角的第一变化量，大于该第一历史特征角相对于上一次计算得到的第二历史特征角的第二变化量。
具体地，在计算得到特征角后，可以将当前计算得到的特征角相对于历史特征角的变化情况，与跌倒判定条件进行比对，在特征角的变化情况满足跌倒判定条件的情况下，得到目标对象发生跌倒的跌倒检测结果。
进一步地，在得到跌倒检测结果后，毫米波雷达还可以以图片、文字和语音等多种形式中的至少一种，或者结合上述至少两种形式，输出该跌倒检测结果，并且该跌倒检测结果的输出对象，可以是存储装置、显示装置或通信装置。此外，毫米波雷达还可以通过通信装置将跌倒检测结果输出至终端。毫米波雷达还可以在跌倒检测结果为发生跌倒的情况下，输出警示信息。
上述对象跌倒检测方法，先获取毫米波雷达发射的电磁波被物体反射后形成的回波信号，再对该回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据，并根据点云数据中的位置信息和信号强度信息，对该点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置，然后基于该质心位置，计算该质心相对于特征点的特征角，最后根据该特征角的变化情况确定目标对象的跌倒检测结果。通过对回波信号进行傅里叶变换处理和聚类分析可以识别出目标对象的点云数据，对该点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置，可以排除环境中目标对象以外其他物体的干扰，并且整个跌倒检测过程无需目标对象佩戴任何辅助传感装置，环境适应性强，有利于提高目标对象的跌倒检测结果的准确性。此外，基于毫米波雷达进行跌倒检测，无隐私泄露风险，且无需佩戴，环境适应性强，有利于扩展跌倒检测方法的应用场景。
需要说明的是，从点云数据中筛选得到目标对象的点云数据的具体方式并不唯一。在一个实施例中，步骤S203包括：结合物体中的目标对象的属性特征或运动特征中的至少一种，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据。
其中，反射电磁波形成回波信号的物体中，包括待进行跌倒检测的目标对象，基于被物体反射形成的回波信号获得目标对象的点云数据的过程，实际上是筛选得到目标对象的点云数据的过程。
具体地，由于不同目标物体的材料和反射面纹路等属性特征不同，对电磁波的反射率以及反射角度也会不同，因此可以根据目标对象的属性特征选择对应的聚类算法实现点云数据的筛选。同样的，由于不同的目标物体的运动特征不同，对应点云数据的速度信息和位置信息的变化情况也会不同，因此也可以根据目标对象的运动特征进行点云数据的筛选。例如家具、墙壁等通常不会发生运动，电风扇等电器会严格按照预设周期运动。基于此，毫米波雷达可以结合物体中的目标对象的属性特征或运动特征中的至少一种，对回波信号进行傅里叶变换处理和聚类分析，筛选得到目标对象的点云数据。
示例性的，毫米波雷达先基于密度聚类算法，将点云数据聚类成多个点云簇，再确定各点云簇的运动特征，将运动特征与目标对象的运动特征匹配的点云簇，确定为包含目标对象的点云数据的点云簇。具体地，基于密度聚类算法，可以聚类得到监测空间中的多个目标物体的点云簇。根据各点云簇的点云数据中的速度信息和位置信息，可以确定各点云簇的运动特征，再将运动特征与目标对象的运动特征进行匹配，即可确定目标对象的点云簇。该点云簇中所包含的点云数据，即为目标对象的点云数据。
上述实施例中，综合考虑的目标对象的属性特征或运动特征中的至少一种，筛选得到目标对象的点云数据，有利于提高点云数据筛选结果的准确性，进而提升跌倒检测结果的准确性。
在一个实施例中，结合物体中的目标对象的属性特征或运动特征中的至少一种，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据，包括：对回波信号进行快速傅里叶变换处理，获得物体的位置信息；根据回波信号的信噪比，确定各位置的信号强度信息；结合物体中的目标对象的属性特征或运动特征中的至少一种，对包含位置信息和信号强度信息的点云数据进行聚类分析，得到目标对象的点云数据。
具体地，毫米波雷达对回波信号进行距离维的快速傅里叶变换，可以得到物体与雷达发射天线的相对距离，进而确定物体的位置信息，并且可以根据各位置所对应的回波信号的信噪比，确定各位置的信号强度信息。可以理解，信噪比越大，对应位置的回波信号能量越大，信号强度也越大。进一步地，毫米波雷达还可以对位置信息进行速度维的快速傅里叶变换，得到物体的速度信息，再将位置信息和速度信息进行合成，生成距离多普勒检测矩阵，并对该距离多普勒检测矩阵进行恒虚警检测，得到多个目标数据，再对目标数据进行快速傅里叶变换，确定角度信息。通过信号处理得到位置和信号强度等信息后，相当于可以得到监测区域的点云数据，该点云数据可以包括位置、信号强度、速度和角度等信息。然后，毫米波雷达再结合物体中的目标对象的属性特征或运动特征中的至少一种，对该点云数据进行聚类分析，即可从点云数据中筛选得到目标对象的点云数据。
本实施例中，通过对回波信号进行快速傅里叶变换处理和聚类分析得到目标对象的点云数据，以减少计算量、提高数据处理效率。
在一些实施例中，步骤S205包括：根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，获得点云数据中每一点云各自的位置和信号强度；基于各点云各自的位置和信号强度，确定各点云所构成的点云簇的质心位置。
其中，信号强度通过信噪比表征。点云数据对应有多个点云，且每一点云各自对应有唯一的位置信息和信号强度信息，点云数据所对应的点云簇，即为点云数据中各点云所构成的点云簇。对点云数据进行信号强度分析确定质心位置的过程，实质上是基于点云数据中各点云各自对应的位置和信号强度，确定各点云所构成的点云簇的质心位置。具体地，毫米波雷达可以根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，获得点云数据中各点云各自对应的位置和信号强度。然后，再基于各点云各自对应的位置和信号强度，确定各点云所构成的点云簇的质心位置。示例性的，可以将信号强度满足强度阈值条件的点云确定为选定点云，并将各选定点云各自位置的平均值或中位数确定为点云簇的质心位置。
上述实施例中，结合位置信息和信号强度信息确定质心位置，可以降低噪声信息的影响，有利于提高质心位置的准确性，进而提升跌倒检测结果的准确性。
在一个实施例中，基于各点云各自的位置和信号强度，确定各点云所构成的点云簇的质心位置，包括：基于各点云各自对应的信号强度信息，确定各点云各自的权重；根据各点云各自的权重，对各点云各自对应的位置信息进行加权求和，确定点云所构成的点云簇的质心位置。
如前文所述，每一个点云对应有一个点云数据，各点云数据中均包含有位置信息和信号强度信息。具体地，毫米波雷达基于各点云各自对应的信号强度信息，可以确定各点云各自的权重，再根据各点云各自的权重，对各点云数据各自对应的位置信息进行加权求和，可以确定点云数据所构成的点云簇的质心位置。
进一步地，确定各点云各自的权重的具体方式并不唯一。例如，在某一点云的信号强度不满足强度阈值条件的情况下，可以将该点云的权重确定为零，并将其余满足强度阈值条件的点云的权重设置为相同值；又如，可以建立信号强度区间与权重的对应关系，针对每一点云，根据该点云的信号强度所在的信号强度区间，确定该点云的权重。
上述实施例中，基于各点云的信号强度确定点云簇的质心，算法简单，有利于提高数据处理效率。
在一个实施例中，基于各点云各自对应的信号强度信息，确定各点云各自的权重，包括：基于各点云各自的信号强度，确定各点云所构成的点云簇的总能量；针对每一点云，将该点云的信号强度与该总能量的比值，确定为该点云的权重。
其中，各点云所构成的点云簇的总能量，为各点云的信号强度之和。具体地，毫米波雷达叠加各点云各自的信号强度，得到各点云所构成的点云簇的总能量，然后，针对每一点云，将该点云的信号强度与该总能量的比值，确定为该点云的权重。
本实施例中，基于各点云的能量分布确定点云的权重，可以自动过滤掉能量较小的点云，有利于提高质心的位置信息的准确性。
在一个实施例中，步骤S209包括：在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒。
其中，跌倒判定条件为各特征角与上一帧的历史特征角的差值中正值或负值的比例大于设定比例，或者，各差值相对于前一差值的差值变化量增大。时间帧是指特征角所对应的时间信息。由于毫米波雷达在进行跌倒检测的过程中，按照一定的周期向外发射电磁波，而特征角是基于某一周期的电磁波的回波信号计算得到的，因此，特征角对应有唯一的时间帧，该时间帧与电磁波的发射时间对应。
具体地，毫米波雷达在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒；在设定数量的连续时间帧内特征角的变化情况不满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为未发生跌倒。
实际场景下，如图3所示，以特征点为雷达坐标原点O，特征角为角OAB的情况为例。若目标对象远离O点，由于目标对象与雷达的距离、以及目标对象的衣服褶皱和姿态等均发生变化，雷达发射的电磁波经目标对象反射后形成的回波信号也将发生变化，反映在点云数据上为：即使不考虑姿态变化的情况下，各点云的信号强度分布也将发生变化。那么，基于各点云各自的信号强度确定的质心位置也将发生变化，也即，在目标对象的平移过程中，目标对象的质心位置并不会维持在平面CAD上，平移对特征角的影响将是波动且无规律的。在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒，且该跌倒判定条件为各特征角与上一帧的历史特征角的差值中正值或负值的比例大于设定比例，或者，各差值相对于前一差值的差值变化量增大，可以体现跌倒过程中特征角的持续有规律的变化过程，能够避免目标对象平移所导致的跌倒误判。
在一个实施例中，在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒，包括：获取设定数量的连续时间帧内计算得到的特征角；基于各特征角各自对应的时间帧，在特征角存在上一帧的历史特征角的情况下，计算特征角与上一帧的历史特征角的差值；在各差值满足跌倒判定条件的情况下，得到目标对象发生跌倒的检测结果。
具体地，毫米波雷达获取设定数量的连续时间帧内计算得到的特征角。该设定数量的具体数值并不唯一，例如可以是10、12、14、15或18等等。然后，毫米波雷达基于各特征角各自对应的时间帧，在特征角存在上一帧的历史特征角的情况下，将特征角减去上一帧的历史特征角，计算得到特征角与上一帧的历史特征角的差值。可以理解，在设定数量为12的情况下，将得到11个差值。最后，毫米波雷达再将各差值与跌倒判定条件进行比对，在各差值满足跌倒判定条件的情况下，得到目标对象发生跌倒的检测结果。以图3为例，特征角为∠OAB的情况下，该跌倒判定条件，可以是各差值中正值的比例大于设定比例，例如，在设定数量为12的情况下，得到11个差值均为正值；也可以是各差值相对于前一差值的差值变化量增大，例如，在设定数量为12的情况下，将得到11个差值再求差，可以得到10个差值变化量，各差值变化量均为正值。
上述实施例中，基于设定数量的连续时间帧内计算得到的特征角进行跌倒判断，有利于提高跌倒检测结果的准确性。
在一个实施例中，如图5所示，对象跌倒检测方法包括：
步骤S501，获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；
步骤S502，对回波信号进行快速傅里叶变换处理，获得物体的位置信息；
步骤S503，根据回波信号的信噪比，确定各位置的信号强度信息；
步骤S504，结合物体中的目标对象的属性特征或运动特征中的至少一种，对包含位置信息和信号强度信息的点云数据进行聚类分析，得到目标对象的点云数据；
步骤S505，根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，获得点云数据中每一点云各自的位置和信号强度；
步骤S506，基于各点云各自的信号强度，确定各点云所构成的点云簇的总能量；
步骤S507，针对每一点云，将该点云的信号强度与总能量的比值，确定为该点云的权重；
步骤S508，根据各点云各自的权重，对各点云各自的位置进行加权求和，确定各点云所构成的点云簇的质心位置；
步骤S509，基于质心位置，计算质心相对于特征点的特征角；
步骤S510，获取设定数量的连续时间帧内计算得到的特征角；
步骤S511，基于各特征角各自对应的时间帧，在特征角存在上一帧的历史特征角的情况下，计算该特征角与上一帧的历史特征角的差值；
步骤S512，在各差值满足跌倒判定条件的情况下，得到目标对象发生跌倒的检测结果。
为便于理解，下面结合图6和图7，对对象跌倒检测方法的具体过程进行详细说明。
在一个实施例中，该毫米波雷达安装在距离地面1.5米高度的墙面，雷达法线方向平行于地面。如图6所示，毫米波雷达通过多进多出(multiple-in multipleout，MIMO)射频收发天线向监测区域发射毫米波段的射频信号，同时接收该射频信号经由监测区域内的障碍物散射得到的回波信号。该回波信号与发射信号混频后形成中频信号，该中频信号经过ADC(Analog to Digital Converter，模数转换)采样后得到采样信息。通过对该采样信息进行距离维的快速傅里叶转换(1D-FFT)，可以获得目标物体的位置信息。对该位置信息进行doppler处理(2D-FFT)，可以获得目标物体的速度信息。将各个通道上的位置信息及速度信息进行合成，生成距离多普勒检测矩阵，然后再对该检测矩阵进行恒虚警(Constant False-Alarm Rate，CFAR)处理和角度计算，过滤掉虚假目标，得到目标物体的点云数据集。该点云数据集包括位置信息、角度信息、多普勒速度信息和信号强度信息。该位置信息通过雷达三维坐标系的坐标值表征，该信号强度信息通过信噪比表征。例如，点云数据集可以表示成pointclouds＝{P0,P1……,Pn-1}。其中，Pi＝[Xi,Yi,Zi,Vi,Si]表示一个点云数据，(Xi,Yi,Zi)为点i的三维坐标，Vi为点i的速度，Si为点i的信噪比。
如图7所示，得到目标物体的点云数据集后，再基于DBSCAN聚类算法对该点云数据集进行聚类分析，将人体的点云数据挑选出来，记为p_cluster。p_cluster＝{pc0,pc1,……,pck}，pci＝[Xci,Yci,Zci,Vci,Sci]表示人体的一个点云数据，0≤i≤k。为确保聚类结果的准确性，k需大于聚类的最小点数5。挑选出的人体点云数据将构成一个点云簇，基于点云簇中各点云数据的能量分布，可以计算得到该点云簇的质心位置。
具体地，先计算点云簇的总能量点云簇总能量SP_cluster_all：
然后将点云簇中每个点所对应的信噪比除以总能量，计算得到点云簇中每个点的权重W_cluster：
W_cluster＝{Sc0/SP_cluster_all，Sc1/SP_cluster_all，Sck/SP_cluster_all}＝{Wc0，......，Wck}      (3)
最后，基于各点各自对应的权重，对各点的坐标信息进行加权求和，计算得到点云簇的质心坐标Ccenter，该质心坐标可以用于表征质心位置。
Ccenter＝[Xc0Wc0+Xc1Wc1+......+XckWck，Yc0Wc0+Yc1Wc1+......
+YckWck，Zc0Wc0+Zc1Wc1+......+ZckWck]       (4)
得到质心坐标后，再根据式(1)计算点云簇的质心相对于毫米波雷达的坐标原点的俯仰角(例如图3中的∠OAB)，并将俯仰角计算结果保存在滑窗w1中，窗体长度为50，w1数组初始值全为0。即w1＝{θ1,0,……,0}，w1数组初始值全为0。
至此，可以得到一个时间帧所对应的俯仰角。
对每一帧重复上述步骤，可以得到连续50帧的俯仰角，存满w1数组，此时，w1数组更新为w1＝{θ1,θ2,……,θ50}。然后，计算w1窗中相邻元素的差α，并将差保存在窗体w2中，w2同样为滑窗保存,窗体长度为49。即w2＝{α1,α2,……,α49}；α1＝θ2-θ1，α2＝θ3-θ2，α49＝θ50-θ49。最后，计算w2窗中相邻元素的差β，并将差保存在窗体w3中，窗体w3同样为滑窗保存,窗体长度为48。即w3＝{β1,β2,……,β48}；β1＝α2-α1，β2＝α3-α2，α48＝α49-α48。若w3中连续不少于10个元素的值为正，则判断该目标对象，即人体发生跌倒。
应该理解的是，虽然如上所述的各实施例所涉及的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，如上所述的各实施例所涉及的流程图中的至少一部分步骤可以包括多个步骤或者多个阶段，这些步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤中的步骤或者阶段的至少一部分轮流或者交替地执行。
基于同样的发明构思，本申请实施例还提供了一种用于实现上述所涉及的对象跌倒检测方法的对象跌倒检测装置。该装置所提供的解决问题的实现方案与上述方法中所记载的实现方案相似，故下面所提供的一个或多个对象跌倒检测装置实施例中的具体限定可以参见上文中对于对象跌倒检测方法的限定，在此不再赘述。
在一个实施例中，如图8所示，提供了一种对象跌倒检测装置800，包括：回波信号获取模块801、目标对象回波信号分析模块802、质心位置确定模块803、特征角计算模块804和跌倒检测结果确定模块805，其中：
回波信号获取模块801，用于获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；
回波信号分析模块802，用于对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；
质心位置确定模块803，用于根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，确定点云数据所对应的点云簇的质心位置；
特征角计算模块804，用于基于质心位置，计算质心相对于特征点的特征角；
跌倒检测结果确定模块805，用于根据特征角的变化情况确定目标对象的跌倒检测结果。
在一个实施例中，质心位置确定模块803具体用于：
根据点云数据中的位置信息和信号强度信息，对点云数据进行信号强度分析，获得点云数据中每一点云各自的位置和信号强度；基于各点云各自的位置和信号强度，确定各点云所构成的点云簇的质心位置；信号强度通过信噪比表征。
在一个实施例中，质心位置确定模块803包括：
权重确定单元，用于基于各点云各自的信号强度，确定各点云各自的权重；及
位置信息确定单元，用于根据各点云各自的权重，对各点云各自的位置进行加权求和，确定各点云所构成的点云簇的质心位置。
在一个实施例中，权重确定单元具体用于：
基于各点云各自的信号强度，确定各点云所构成的点云簇的总能量；针对每一点云，将该点云的信号强度与总能量的比值，确定为该点云的权重。
在一个实施例中，回波信号分析模块802具体用于：
结合物体中的目标对象的属性特征或运动特征中的至少一种，对回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据。
在一个实施例中，回波信号分析模块802具体用于：
对回波信号进行快速傅里叶变换处理，获得物体的位置信息；根据回波信号的信噪比，确定各位置的信号强度信息；结合物体中的目标对象的属性特征或运动特征中的至少一种，对包含位置信息和信号强度信息的点云数据进行聚类分析，得到目标对象的点云数据。
在一个实施例中，特征点为雷达坐标原点；特征角的计算公式为：
式中，θ为特征角，x、y和z共同表征质心位置。
在一个实施例中，跌倒检测结果确定模块805具体用于：
在设定数量的连续时间帧内特征角的变化情况满足跌倒判定条件的情况下，确定目标对象的跌倒检测结果为发生跌倒；跌倒判定条件为各特征角与上一帧的历史特征角的差值中正值或负值的比例大于设定比例，或者，各差值相对于前一差值的差值变化量增大。
在一个实施例中，跌倒检测结果确定模块805具体用于：
获取设定数量的连续时间帧内计算得到的特征角；基于各特征角各自对应的时间帧，在特征角存在上一帧的历史特征角的情况下，计算该特征角与上一帧的历史特征角的差值；在各差值满足跌倒判定条件的情况下，得到目标对象发生跌倒的检测结果。
上述对象跌倒检测装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于毫米波雷达中的处理器中，也可以以软件形式存储于毫米波雷达中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中，如图9所示，提供了一种毫米波雷达，该毫米波雷达包括雷达信号收发装置901和处理器902。该雷达信号收发装置901用于发射电磁波，并接收该电磁波被物体反射后形成的回波信号；该处理器902用于实现上述对象跌倒检测方法中的步骤。
其中，雷达信号收发装置901可以包括发射装置和接收装置，也可以是收发一体的集成装置。进一步地，该雷达信号收发装置901还可以是射频收发装置。该处理器902可以是包含各类处理芯片及其外围电路，具备逻辑运算功能的硬件模块。该处理芯片，可以是单片机、DSP(Digital Signal Process，数字信号处理)芯片或FPGA(Field Programmable Gate Array，现场可编程逻辑门阵列)芯片。关于对象跌倒检测方法的具体限定参见上文，此处不再赘述。具体地，该雷达信号收发装置901用于发射电磁波，并接收该电磁波被物体反射后形成的回波信号；该处理器902用于实现上述对象跌倒检测方法中的步骤。
进一步地，如图9所示，该毫米波雷达还可以包括通信装置903和电源装置904。通信装置903可以是有线通信装置或无线通信装置。该有线通信装置，可以为总线通信装置，如485通信装置、CAN通信装置或RS232通信装置。该无线通信装置，可以是蓝牙通信装置、无线通信装置或蜂窝通信装置。该电源装置904可以是供电插头，用于向外部电源获取电能，也可以是包含储能器件，可以向外输出电能的装置。该储能器件，可以是储能电池组或超级电容。处理器902可以通过串口分别连接雷达信号收发装置901和通信装置903。
此外，毫米波雷达还可以包括存储装置、显示装置和警示装置等等。该存储装置用于存储计算机程序，处理器902处理该计算机程序时实现上述的对象跌倒检测方法。该显示装置用于显示跌倒检测结果，该警示装置用于在跌倒检测结果为发生跌倒的情况下输出警示信息。本领域技术人员可以理解，图9中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的毫米波雷达的限定，具体地毫米波雷达可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
在其中一个实施例中，提供了一种计算机可读存储介质，其上存储有计算机程序，计算机程序被处理器执行时实现上述对象跌倒检测方法中的步骤。
在其中一个实施例中，提供了一种计算机程序产品，包括计算机程序，该计算机程序被处理器执行时实现上述对象跌倒检测方法中的步骤。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、数据库或其它介质的任何引用，均可包括非易失性和易失性存储器中的至少一种。非易失性存储器可包括只读存储器(Read-Only Memory，ROM)、磁带、软盘、闪存、光存储器、高密度嵌入式非易失性存储器、阻变存储器(ReRAM)、磁变存储器(Magnetoresistive Random Access Memory，MRAM)、铁电存储器(Ferroelectric Random Access Memory，FRAM)、相变存储器(Phase Change Memory，PCM)、石墨烯存储器等。易失性存储器可包括随机存取存储器(Random Access Memory，RAM)或外部高速缓冲存储器等。作为说明而非局限，RAM可以是多种形式，比如静态随机存取存储器(Static Random Access Memory，SRAM)或动态随机存取存储器(Dynamic Random Access Memory，DRAM)等。本申请所提供的各实施例中所涉及的数据库可包括关系型数据库和非关系型数据库中至少一种。非关系型数据库可包括基于区块链的分布式数据库等，不限于此。本申请所提供的各实施例中所涉及的处理器可为通用处理器、中央处理器、图形处理器、数字信号处理器、可编程逻辑器、基于量子计算的数据处理逻辑器等，不限于此。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对本申请专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种对象跌倒检测方法，其特征在于，所述方法包括：
    获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；
    对所述回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；
    根据所述点云数据中的位置信息和信号强度信息，对所述点云数据进行信号强度分析，确定所述点云数据所对应的点云簇的质心位置；
    基于所述质心位置，计算质心相对于特征点的特征角；及
    根据所述特征角的变化情况确定所述目标对象的跌倒检测结果。
  2. 根据权利要求1所述的方法，其特征在于，所述根据所述点云数据中的位置信息和信号强度信息，对所述点云数据进行信号强度分析，确定所述点云数据所对应的点云簇的质心位置，包括：
    根据所述点云数据中的位置信息和信号强度信息，对所述点云数据进行信号强度分析，获得所述点云数据中每一点云各自的位置和信号强度；及
    基于各所述点云各自的位置和信号强度，确定各所述点云所构成的点云簇的质心位置；所述信号强度通过信噪比表征。
  3. 根据权利要求2所述的方法，其特征在于，所述基于各所述点云各自的位置和信号强度，确定各所述点云所构成的点云簇的质心位置，包括：
    基于各所述点云各自的信号强度，确定各所述点云各自的权重；及
    根据各所述点云各自的权重，对各所述点云各自的位置进行加权求和，确定各所述点云所构成的点云簇的质心位置。
  4. 根据权利要求3所述的方法，其特征在于，所述基于各所述点云各自的信号强度，确定各所述点云各自的权重，包括：
    基于各所述点云各自的信号强度，确定各所述点云所构成的点云簇的总能量；及
    针对每一所述点云，将所述点云的信号强度与所述总能量的比值，确定为所述点云的权重。
  5. 根据权利要求1至4中任意一项所述的方法，其特征在于，所述对所述回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据，包括：
    结合所述物体中的目标对象的属性特征或运动特征中的至少一种，对所述回波信号进行傅里叶变换处理和聚类分析，得到所述目标对象的点云数据。
  6. 根据权利要求5所述的方法，其特征在于，所述结合所述物体中的目标对象的属性特征或运动特征中的至少一种，对所述回波信号进行傅里叶变换处理和聚类分析，得到所述目标对象的点云数据，包括：
    对所述回波信号进行快速傅里叶变换处理，获得所述物体的位置信息；
    根据所述回波信号的信噪比，确定各位置的信号强度信息；及
    结合所述物体中的目标对象的属性特征或运动特征中的至少一种，对包含所述位置信息和所述信号强度信息的点云数据进行聚类分析，得到所述目标对象的点云数据。
  7. 根据权利要求1至4中任意一项所述的方法，其特征在于，所述特征点为雷达坐标原点；所述特征角的计算公式为：
    式中，θ为特征角，x、y和z共同表征所述质心位置。
  8. 根据权利要求1至7中任意一项所述的方法，其特征在于，所述根据所述特征角的变化情况确定所述目标对象的跌倒检测结果，包括：
    在设定数量的连续时间帧内所述特征角的变化情况满足跌倒判定条件的情况下，确定所述目标对象的跌倒检测结果为发生跌倒；所述跌倒判定条件为各所述特征角与上一帧的历史特征角的差值中正值或负值的比例大于设定比例，或者，各所述差值相对于前一差值的差值变化量增大。
  9. 根据权利要求8所述的方法，其特征在于，所述在设定数量的连续时间帧内所述特征角的变化情况满足跌倒判定条件的情况下，确定所述目标对象的跌倒检测结果为发生跌倒，包括：
    获取设定数量的连续时间帧内计算得到的所述特征角；
    基于各所述特征角各自对应的时间帧，在所述特征角存在上一帧的历史特征角的情况下，计算所述特征角与上一帧的历史特征角的差值；及
    在各所述差值满足跌倒判定条件的情况下，得到所述目标对象发生跌倒的检测结果。
  10. 一种对象跌倒检测装置，其特征在于，所述装置包括：
    回波信号获取模块，用于获取毫米波雷达发射的电磁波被物体反射后形成的回波信号；
    回波信号分析模块，用于对所述回波信号进行傅里叶变换处理和聚类分析，得到目标对象的点云数据；
    质心位置确定模块，用于根据所述点云数据中的位置信息和信号强度信息，对所述点云数据进行信号强度分析，确定所述点云数据所对应的点云簇的质心位置；
    特征角计算模块，用于基于所述质心位置，计算质心相对于特征点的特征角；及
    跌倒检测结果确定模块，用于根据所述特征角的变化情况确定所述目标对象的跌倒检测结果。
  11. 根据权利要求10所述的装置，其特征在于，所述质心位置确定模块具体用于：
    根据所述点云数据中的位置信息和信号强度信息，对所述点云数据进行信号强度分析，获得所述点云数据中每一点云各自的位置和信号强度；及
    基于各所述点云各自的位置和信号强度，确定各所述点云所构成的点云簇的质心位置；所述信号强度通过信噪比表征。
  12. 根据权利要求11所述的装置，其特征在于，所述质心位置确定模块包括：
    权重确定单元，用于基于各所述点云各自的信号强度，确定各所述点云各自的权重；及
    位置确定单元，用于根据各所述点云各自的权重，对各所述点云各自的位置进行加权求和，确定各所述点云所构成的点云簇的质心位置。
  13. 根据权利要求12所述的装置，其特征在于，所述权重确定单元具体用于：
    基于各所述点云各自的信号强度，确定各所述点云所构成的点云簇的总能量；及
    针对每一所述点云，将所述点云的信号强度与所述总能量的比值，确定为所述点云的权重。
  14. 根据权利要求10至13中任意一项所述的装置，其特征在于，所述回波信号分析模块具体用于：
    结合所述物体中的目标对象的属性特征或运动特征中的至少一种，对所述回波信号进行傅里叶变换处理和聚类分析，得到所述目标对象的点云数据。
  15. 根据权利要求14所述的装置，其特征在于，所述回波信号分析模块具体用于：
    对所述回波信号进行快速傅里叶变换处理，获得所述物体的位置信息；
    根据所述回波信号的信噪比，确定各位置的信号强度信息；及
    结合所述物体中的目标对象的属性特征或运动特征中的至少一种，对包含所述位置信息和所述信号强度信息的点云数据进行聚类分析，得到所述目标对象的点云数据。
  16. 根据权利要求10至15中任意一项所述的装置，其特征在于，所述跌倒检测结果确定模块具体用于：
    在设定数量的连续时间帧内所述特征角的变化情况满足跌倒判定条件的情况下，确定所述目标对象的跌倒检测结果为发生跌倒；所述跌倒判定条件为各所述特征角与上一帧的历史特征角的差值中正值或负值的比例大于设定比例，或者，各所述差值相对于前一差值的差值变化量增大。
  17. 根据权利要求16所述的装置，其特征在于，所述跌倒检测结果确定模块具体用于：
    获取设定数量的连续时间帧内计算得到的所述特征角；
    基于各所述特征角各自对应的时间帧，在所述特征角存在上一帧的历史特征角的情况下，计算所述特征角与上一帧的历史特征角的差值；及
    在各所述差值满足跌倒判定条件的情况下，得到所述目标对象发生跌倒的检测结果。
  18. 一种毫米波雷达，其特征在于，包括雷达信号收发装置和处理器，所述雷达信号收发装置用于发射电磁波，并接收所述电磁波被物体反射后形成的回波信号；所述处理器用于实现权利要求1至9中任一项所述的方法的步骤。
  19. 一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，其特征在于，所述计算机可读指令被一个或多个处理器执行时，使得所述一个或多个处理器实现权利要求1至9中任一项所述的方法的步骤。
  20. 一种计算机程序产品，其特征在于，包括计算机程序，所述计算机程序被处理器执行时实现权利要求1至9中任一项所述的方法的步骤。
PCT/CN2023/098845 2022-06-10 2023-06-07 对象跌倒检测方法、装置、毫米波雷达和存储介质 WO2023236989A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210651286.4A CN114720977B (zh) 2022-06-10 2022-06-10 对象跌倒检测方法、装置、毫米波雷达和存储介质
CN202210651286.4 2022-06-10

Publications (1)

Publication Number Publication Date
WO2023236989A1 true WO2023236989A1 (zh) 2023-12-14

Family

ID=82233082

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/098845 WO2023236989A1 (zh) 2022-06-10 2023-06-07 对象跌倒检测方法、装置、毫米波雷达和存储介质

Country Status (2)

Country Link
CN (1) CN114720977B (zh)
WO (1) WO2023236989A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117647788A (zh) * 2024-01-29 2024-03-05 北京清雷科技有限公司 一种基于人体3d点云的危险行为识别方法及设备
CN117831224A (zh) * 2024-02-29 2024-04-05 深圳市迈远科技有限公司 基于毫米雷达的跌倒报警方法、装置、设备及介质

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114720977B (zh) * 2022-06-10 2022-09-20 长沙莫之比智能科技有限公司 对象跌倒检测方法、装置、毫米波雷达和存储介质
CN116106855B (zh) * 2023-04-13 2023-07-18 中国科学技术大学 摔倒检测方法及摔倒检测装置
CN116338684A (zh) * 2023-05-29 2023-06-27 长沙莫之比智能科技有限公司 基于毫米波雷达与深度学习的人体跌倒检测方法及系统
CN116439694B (zh) * 2023-06-14 2023-08-15 深圳市魔样科技有限公司 基于运动模型训练的智能手表动态数据监测方法
CN117636404B (zh) * 2024-01-26 2024-04-16 贵州信邦富顿科技有限公司 基于非穿戴式设备的跌倒检测方法及系统

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2017151043A (ja) * 2016-02-26 2017-08-31 株式会社デンソー 物体認識装置及びプログラム
US20190099113A1 (en) * 2016-01-07 2019-04-04 Gunther Röder Method and device for detecting a fall
CN111191499A (zh) * 2019-11-11 2020-05-22 广东工业大学 一种基于最小中心线的跌倒检测方法及装置
CN112386248A (zh) * 2019-08-13 2021-02-23 中国移动通信有限公司研究院 人体跌倒的检测方法、装置、设备及计算机可读存储介质
CN113392681A (zh) * 2020-03-13 2021-09-14 深圳云天励飞技术有限公司 人体跌倒检测方法、装置及终端设备
CN114332922A (zh) * 2021-12-16 2022-04-12 上海电机学院 一种基于图像静态特征的摔倒检测方法
CN114720977A (zh) * 2022-06-10 2022-07-08 长沙莫之比智能科技有限公司 对象跌倒检测方法、装置、毫米波雷达和存储介质

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
FR3028343B1 (fr) * 2014-11-10 2016-12-23 Centre Nat Rech Scient Procede de detection de chute d'un sujet humain et dispositif actimetrique correspondant
CN111134685B (zh) * 2018-11-02 2022-08-09 富士通株式会社 跌倒检测方法和装置
CN111166342B (zh) * 2020-01-07 2022-10-14 四川宇然智荟科技有限公司 一种毫米波雷达和摄像头融合跌倒检测装置及其检测方法
CN114005246B (zh) * 2021-01-29 2024-01-30 江苏中科西北星信息科技有限公司 一种基于调频连续波毫米波雷达的老人跌倒检测方法及装置
CN114220119B (zh) * 2021-11-10 2022-08-12 深圳前海鹏影数字软件运营有限公司 人体姿态检测方法、终端设备及计算机可读存储介质
CN114330726A (zh) * 2021-12-29 2022-04-12 广东电网有限责任公司 追踪定位方法、装置、电子设备及存储介质

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190099113A1 (en) * 2016-01-07 2019-04-04 Gunther Röder Method and device for detecting a fall
JP2017151043A (ja) * 2016-02-26 2017-08-31 株式会社デンソー 物体認識装置及びプログラム
CN112386248A (zh) * 2019-08-13 2021-02-23 中国移动通信有限公司研究院 人体跌倒的检测方法、装置、设备及计算机可读存储介质
CN111191499A (zh) * 2019-11-11 2020-05-22 广东工业大学 一种基于最小中心线的跌倒检测方法及装置
CN113392681A (zh) * 2020-03-13 2021-09-14 深圳云天励飞技术有限公司 人体跌倒检测方法、装置及终端设备
CN114332922A (zh) * 2021-12-16 2022-04-12 上海电机学院 一种基于图像静态特征的摔倒检测方法
CN114720977A (zh) * 2022-06-10 2022-07-08 长沙莫之比智能科技有限公司 对象跌倒检测方法、装置、毫米波雷达和存储介质

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117647788A (zh) * 2024-01-29 2024-03-05 北京清雷科技有限公司 一种基于人体3d点云的危险行为识别方法及设备
CN117647788B (zh) * 2024-01-29 2024-04-26 北京清雷科技有限公司 一种基于人体3d点云的危险行为识别方法及设备
CN117831224A (zh) * 2024-02-29 2024-04-05 深圳市迈远科技有限公司 基于毫米雷达的跌倒报警方法、装置、设备及介质
CN117831224B (zh) * 2024-02-29 2024-05-24 深圳市迈远科技有限公司 基于毫米雷达的跌倒报警方法、装置、设备及介质

Also Published As

Publication number Publication date
CN114720977A (zh) 2022-07-08
CN114720977B (zh) 2022-09-20

Similar Documents

Publication Publication Date Title
WO2023236989A1 (zh) 对象跌倒检测方法、装置、毫米波雷达和存储介质
JP2020071226A (ja) 転倒検出方法及び装置
TW202009650A (zh) 促進使用者與擴增實境介面中之顯示物件互動之簡易性及準確性之基於智慧型手機之雷達系統
CN111142102B (zh) 一种呼吸数据计算方法以及相关设备
CN110032670A (zh) 时序数据的异常检测方法、装置、设备及存储介质
WO2011055718A1 (ja) 測位装置、及びこれを用いたセンサ情報の統合解析による見守りシステム
EP3342330B1 (en) Phasor approach to signal to noise ratio measurement evaluation of physiological signals
Gelb et al. Statistics of distinct clutter classes in midfrequency active sonar
CN113885022A (zh) 一种跌倒检测方法及雷达设备
US20220326370A1 (en) Distance measurement method and distance measurement apparatus
KR102524373B1 (ko) 실내 보안 응용을 위한 인체 감지 레이다 시스템 및 이를 이용한 인체 감지 방법
CN113311428A (zh) 一种基于毫米波雷达的人体跌倒智能监测系统及识别方法
CN115755015A (zh) 座舱内的活体检测方法、装置、设备及介质
CN116087943A (zh) 基于毫米波雷达的室内跌倒检测方法及系统
Zhang et al. Application of multi-angle millimeter-wave radar detection in human motion behavior and micro-action recognition
Xie et al. mmEat: Millimeter wave-enabled environment-invariant eating behavior monitoring
CN110784837A (zh) 室内定位方法、装置、介质及电子设备
CN113326449B (zh) 预测交通流量的方法、装置、电子设备和介质
Gálvez et al. Improved neural network based CFAR detection for non homogeneous background and multiple target situations
JP7484492B2 (ja) レーダーに基づく姿勢認識装置、方法及び電子機器
CN113420634A (zh) 人体摔倒的检测方法、装置、电子设备及存储介质
WO2023116455A1 (zh) 极化雷达的广义相似性度量方法、装置、设备及存储介质
CN116236173B (zh) 基于毫米波雷达的智能养老监护管理系统
Qiao et al. Radar point clouds processing for human activity classification using convolutional multilinear subspace learning
WO2021092771A1 (zh) 一种目标检测方法及装置、设备、存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23819154

Country of ref document: EP

Kind code of ref document: A1