CN117281498B - Health risk early warning method and equipment based on millimeter wave radar - Google Patents
Health risk early warning method and equipment based on millimeter wave radar Download PDFInfo
- Publication number
- CN117281498B CN117281498B CN202311577651.2A CN202311577651A CN117281498B CN 117281498 B CN117281498 B CN 117281498B CN 202311577651 A CN202311577651 A CN 202311577651A CN 117281498 B CN117281498 B CN 117281498B
- Authority
- CN
- China
- Prior art keywords
- radar
- moving
- moving object
- doppler
- range profile
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 45
- 230000036541 health Effects 0.000 title claims abstract description 34
- 239000011159 matrix material Substances 0.000 claims abstract description 49
- 238000000605 extraction Methods 0.000 claims abstract description 46
- 238000012544 monitoring process Methods 0.000 claims abstract description 44
- 238000004458 analytical method Methods 0.000 claims abstract description 29
- 230000009471 action Effects 0.000 claims abstract description 25
- 238000007781 pre-processing Methods 0.000 claims abstract description 9
- 230000006399 behavior Effects 0.000 claims description 68
- 230000000875 corresponding effect Effects 0.000 claims description 52
- 238000001514 detection method Methods 0.000 claims description 32
- 238000012545 processing Methods 0.000 claims description 16
- 238000010586 diagram Methods 0.000 claims description 13
- 238000000513 principal component analysis Methods 0.000 claims description 13
- 238000009826 distribution Methods 0.000 claims description 7
- 238000011065 in-situ storage Methods 0.000 claims description 7
- 239000013598 vector Substances 0.000 claims description 7
- 238000010606 normalization Methods 0.000 claims description 2
- 238000013473 artificial intelligence Methods 0.000 abstract description 2
- 238000004590 computer program Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 6
- 239000006185 dispersion Substances 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000004422 calculation algorithm Methods 0.000 description 4
- 238000012935 Averaging Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 238000005406 washing Methods 0.000 description 3
- 230000007423 decrease Effects 0.000 description 2
- 230000001419 dependent effect Effects 0.000 description 2
- 201000010099 disease Diseases 0.000 description 2
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 230000003068 static effect Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 208000030090 Acute Disease Diseases 0.000 description 1
- 206010010774 Constipation Diseases 0.000 description 1
- 208000013738 Sleep Initiation and Maintenance disease Diseases 0.000 description 1
- 230000032683 aging Effects 0.000 description 1
- 230000003321 amplification Effects 0.000 description 1
- 238000005452 bending Methods 0.000 description 1
- 230000037237 body shape Effects 0.000 description 1
- 230000001680 brushing effect Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 208000002173 dizziness Diseases 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 206010022437 insomnia Diseases 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000027939 micturition Effects 0.000 description 1
- 238000003199 nucleic acid amplification method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000012847 principal component analysis method Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000004622 sleep time Effects 0.000 description 1
- 230000036578 sleeping time Effects 0.000 description 1
- 210000002784 stomach Anatomy 0.000 description 1
- 230000001131 transforming effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
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
- G01S13/886—Radar or analogous systems specially adapted for specific applications for alarm systems
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/05—Detecting, measuring or recording for diagnosis by means of electric currents or magnetic fields; Measuring using microwaves or radio waves
-
- 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
-
- 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
- A61B5/1123—Discriminating type of movement, e.g. walking or running
-
- 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
- A61B5/7271—Specific aspects of physiological measurement analysis
- A61B5/7275—Determining trends in physiological measurement data; Predicting development of a medical condition based on physiological measurements, e.g. determining a risk factor
-
- 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
- G01S13/02—Systems using reflection of radio waves, e.g. primary radar systems; Analogous systems
- G01S13/04—Systems determining presence of a target
-
- 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
-
- 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
- G01S7/415—Identification of targets based on measurements of movement associated with the target
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/762—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using clustering, e.g. of similar faces in social networks
- G06V10/763—Non-hierarchical techniques, e.g. based on statistics of modelling distributions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/7715—Feature extraction, e.g. by transforming the feature space, e.g. multi-dimensional scaling [MDS]; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/80—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level
- G06V10/806—Fusion, i.e. combining data from various sources at the sensor level, preprocessing level, feature extraction level or classification level of extracted features
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Remote Sensing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Radar, Positioning & Navigation (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Multimedia (AREA)
- Evolutionary Computation (AREA)
- Veterinary Medicine (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- Public Health (AREA)
- Animal Behavior & Ethology (AREA)
- Computer Networks & Wireless Communication (AREA)
- Surgery (AREA)
- Molecular Biology (AREA)
- Heart & Thoracic Surgery (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Pathology (AREA)
- Physiology (AREA)
- Psychiatry (AREA)
- Dentistry (AREA)
- Oral & Maxillofacial Surgery (AREA)
- Electromagnetism (AREA)
- Radiology & Medical Imaging (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Probability & Statistics with Applications (AREA)
- Social Psychology (AREA)
- Human Computer Interaction (AREA)
- Signal Processing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention relates to the technical fields of intelligent medical treatment and artificial intelligence, in particular to a health risk early warning method and equipment based on millimeter wave radar, wherein the method comprises the steps of collecting radar echo signals in a radar monitoring scene and preprocessing; determining the movement direction and movement speed of a target in a radar monitoring scene; judging whether targets exist in the radar monitoring scene according to the multi-scale high-resolution range profile; the feature extraction matrix, the category templates and the labels are updated regularly, feature extraction is carried out in real time by utilizing the feature extraction matrix under the condition that targets exist, real-time feature data are obtained, and the real-time feature data are classified according to the category templates and the behavior labels, so that action types of the targets at the current moment are obtained; and carrying out health risk early warning on the target according to the action type of the target and the radar monitoring scene. The invention realizes the identification and analysis of human behavior actions so as to provide accurate health risk early warning through real-time actions.
Description
Technical Field
The invention relates to the technical fields of intelligent medical treatment and artificial intelligence, in particular to a health risk early warning method and equipment based on millimeter wave radar.
Background
In recent years, the trend of aging population is continuously increased, the population number of the elderly living alone is continuously increased, and the human body can be in a negative state such as a reaction retardation, a slow action, a decline of balance ability and the like due to the decline of the physical function of the elderly, so that the probability of occurrence of accidents such as falling and acute diseases is increased. The emergency situation of the old can be timely and accurately found, the old can be timely sent to a hospital for medical treatment, the cure rate can be effectively improved, and the life safety of the old is ensured.
The existing human behavior recognition method mainly comprises the following steps: human body behavior recognition method based on contact sensor, human body behavior recognition method based on audio sensor, human body behavior recognition method based on visual sensor, and human body behavior recognition method based on radar sensor. The prior art scheme has some defects, and the touch sensor is easy to generate false alarm and detection omission; the noise of the sound information collected by the audio sensor is large, and the accuracy of sound judgment can be affected; the scene limitation of the image acquired by the vision sensor is larger; the radar sensor collects data independent of the environment and scene.
Particularly, the human body behavior detection method based on the radar sensor is usually to monitor in a mode of presetting a threshold value, and different testers have differences in height, body shape and the like, so that the monitoring results tend to be large in difference, and the results are inaccurate.
Disclosure of Invention
In view of the foregoing, in one aspect, the present invention provides a health risk early warning method based on millimeter wave radar, the method comprising:
collecting radar echo signals in a radar monitoring scene;
preprocessing the radar echo signals to obtain a multi-scale high-resolution range profile, a range Doppler image and angles of moving targets relative to the radar;
determining the motion direction and the motion speed of a moving target in a radar monitoring scene according to the range-Doppler diagram;
judging whether a moving target exists in the radar monitoring scene according to the multi-scale high-resolution range profile;
extracting characteristics of the multi-scale high-resolution range profile, the motion direction and the motion speed by using a characteristic extraction matrix, and judging the action type of a moving object at the current moment according to a category template and a corresponding behavior label, wherein the characteristic extraction matrix, the category template and the corresponding behavior label are dynamic data obtained by using the multi-scale high-resolution range profile, the motion direction and the motion speed in a period of time before the current moment;
and carrying out health risk early warning on the moving target according to the action type of the moving target and the radar monitoring scene.
Optionally, the health risk early warning method based on millimeter wave radar provided by the invention further comprises the following steps:
storing the multi-scale high-resolution range profile, the moving direction, the moving speed and the moving target judgment result to obtain a history reference characteristic;
performing principal component analysis on the historical reference features at preset time intervals, and determining a principal component feature extraction matrix based on analysis results；
Extracting matrix by using the principal component featuresExtracting the characteristic of the history reference characteristic to obtain main component characteristic +.>；
For the principal component characteristicsClustering is carried out to obtain a feature classification result;
respectively classifying various data in the results according to the characteristicsCorresponding multiscale high-resolution range profile, movement direction and movement speed determine various data +.>Corresponding behavior tags.
Optionally, principal component analysis is performed on the historical reference features, and a principal component feature extraction matrix is determined based on the analysis resultsComprising:
splicing the multi-scale high-resolution range profile, the motion direction and the motion speed when the moving target exists in the historical reference characteristics within the preset time to obtain characteristic data；
For the characteristic data in the slow time dimensionStandardized processing to obtain standard characteristic data +. >；
To the instituteThe standard characteristic dataPerforming principal component analysis, and determining principal component feature extraction matrix based on analysis result。
Optionally, determining a principal component feature extraction matrix based on the analysis resultsComprising:
according to the standard characteristic dataDetermining principal component feature extraction matrix by using feature value and feature vector obtained by principal component analysis>；
Determining the dimension according to the ratio of the characteristic value of each dimension to the characteristic value of all dimensions；
Extracting matrix from the principal component featuresExtracting maximum->Feature vectors corresponding to the dimensional feature values obtain a principal component feature extraction matrix +.>。
Optionally, determining the dimension according to the ratio of the characteristic value of each dimension to the characteristic value of all the dimensionsComprising:
computing the sum of the eigenvalues of the previous i dimension and the eigenvalues of all dimensionsRatio of sumsThe initial value of i is 1;
judging the ratioWhether or not threshold is reached +.>；
If the ratio isReaching threshold->Then the current i is taken as dimension +.>；
If the ratio isDoes not reach threshold +.>The value of i is increased.
Optionally, classifying various data in the results according to the characteristics respectivelyCorresponding multi-scale high-resolution range profile, motion direction, motion speed, radar height and angle of moving target relative to radar to determine various data A corresponding behavior tag comprising:
extracting the target height at the corresponding moment according to the multi-scale high-resolution range profile, the radar height and the angle;
calculating various dataAverage height of (2);
determining various data according to the average heightObtaining a category template;
and determining the behavior label under each cluster label according to the movement direction and movement speed corresponding to each cluster label.
Optionally, extracting features of the multi-scale high-resolution range profile, the motion direction and the motion speed by using a feature extraction matrix includes:
splicing the multiscale high-resolution range profile, the motion direction and the motion speed when the moving target exists to obtain real-time characteristic data；
For the real-time feature data in the slow time dimensionPerforming standardization processing to obtain real-time standard characteristic data；
Extracting matrix by using the principal component featuresFor the real-time standard characteristic data +.>Extracting features to obtain real-time main component features ∈ ->。
Optionally, preprocessing the radar echo signal to obtain a multi-scale high-resolution range profile, a range-doppler plot, and an angle of a moving target relative to the radar, including:
extracting a distance dimension complex signal of the radar echo signal, and extracting a moving target from the distance dimension complex signal according to a time scale to obtain a fine-scale high-resolution range profile and a coarse-scale high-resolution range profile;
Processing the distance dimension complex signal in a slow time dimension, and extracting a distance Doppler graph of a moving target;
and carrying out conjugate multiplication on the distance dimension complex signals extracted by different radar antennas, and solving an average value to obtain phase-coherent signals in different directions, and extracting angles of moving targets in different directions relative to the radar according to the phase-coherent signals in different directions.
Optionally, determining a motion direction and a motion speed of a moving target in the radar monitoring scene according to the range-doppler plot includes:
performing constant false alarm detection on the range-Doppler graph to obtain a detection threshold;
dividing the range-Doppler graph into a radar area, a radar area and an in-situ stay area according to Doppler values, and respectively calculating energy sums of point clouds in each area, which are higher than a detection threshold, so as to obtain three Doppler energy sums;
comparing the three Doppler energy sums with three motion direction discrimination thresholds to determine the motion direction of the moving target;
and determining the moving speed of the moving object according to the Doppler value of the area corresponding to the moving direction of the moving object.
Optionally, judging whether a moving target exists in the radar monitoring scene according to the multi-scale high-resolution range profile includes:
Calculating the first target existence characteristic according to the probability distribution of the energy value of the coarse-scale high-resolution range profile;
determining the existence characteristic of the second target according to the relativity of the energy value and the energy mean value of the coarse-scale high-resolution range profile;
if the first target existence characteristic is larger than the first unmanned detection threshold value and the second target existence characteristic is larger than the second unmanned detection threshold value, determining that a moving target exists in the radar monitoring scene at the current moment.
Optionally, performing health risk early warning on the moving target according to the action type of the moving target and the radar monitoring scene, including:
counting action types of the moving object at a plurality of moments;
and predicting the action types of the moving target at a plurality of moments according to the radar monitoring scene to obtain a health risk result.
Optionally, the health risk early warning method based on millimeter wave radar provided by the invention further comprises the following steps:
counting behavior labels of each type of moving targets in preset historical time;
and judging the behavior habit of each type of moving object in the corresponding radar monitoring scene according to the statistical behavior label.
In another aspect of the present invention, there is also provided a health risk early warning device based on millimeter wave radar, the device comprising: a processor and a memory coupled to the processor; the memory stores instructions executable by the processor to cause the processor to perform a health risk early warning method based on millimeter wave radar technology.
According to the health risk early warning method and the health risk early warning device based on the millimeter wave radar, the radar echo signals received by radar equipment in a radar monitoring scene are preprocessed, the moving object detection, the behavior analysis and the early warning are carried out, noise in the scene is reduced, the accuracy of signal analysis is improved, the information such as the multiscale high-resolution range profile, the range Doppler diagram and the angle of the moving object relative to the radar obtained through preprocessing can be used for realizing the detection of the moving speed and the moving direction of the moving object, then the real-time tracking of the moving object is realized by judging whether the moving object exists in the radar monitoring scene, the human body behavior analysis is carried out on the moving object in the radar monitoring scene by utilizing the feature extraction matrix, the category template and the behavior label which are updated regularly, the specific and self-adaptive behavior recognition can be realized for the difference of different testers, so that more accurate health risk early warning can be provided, the occurrence risk of unexpected situations such as moving object diseases, falling can be accurately predicted, and unexpected occurrence is reduced.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is an application scenario diagram of a millimeter wave radar device provided in an embodiment of the present invention;
fig. 2 is a schematic flow chart of a health risk early warning method based on millimeter wave radar according to an embodiment of the present invention;
fig. 3 is an example of an antenna arrangement of a millimeter wave radar device provided by an embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention will be made apparent and fully in view of the accompanying drawings, in which some, but not all embodiments of the invention are shown. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc. are based on the directions or positional relationships shown in the drawings, are merely for convenience of describing the present invention and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a specific orientation, be configured and operated in a specific orientation, and thus should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
In addition, the technical features of the different embodiments of the present invention described below may be combined with each other as long as they do not collide with each other.
As shown in fig. 1, in the application scenario diagram of the millimeter wave radar device provided by the embodiment of the invention, the millimeter wave radar device is installed at a proper position of a roof in a radar monitoring scenario, and data in the radar monitoring scenario is collected to perform human behavior recognition and health risk prediction. The radar emits a Frequency Modulation Continuous Wave (FMCW) signal, the FMCW signal in one period is a Chirp signal, the signal modulation mode is saw tooth wave or triangular wave, and the period of the Chirp signal isN Chirp signals transmitted continuously form a frame with a frame period of +.>。
As shown in fig. 2, an embodiment of the present invention provides a health risk early warning method based on millimeter wave radar, including:
s1, acquiring radar echo signals in a radar monitoring scene;
in the embodiment, the radar echo signal is obtained by mixing an echo signal received by millimeter wave radar equipment with a transmitting signal to obtain a difference frequency signal, and obtaining a digitized echo signal through high-pass filtering, low-noise amplification and ADC sampling; the acquired digital echo signals are uploaded to the cloud end in real time through a network.
S2, preprocessing radar echo signals to obtain a multi-scale high-resolution range profile, a range Doppler image and angles of moving targets relative to the radar;
and S3, determining the movement direction and movement speed of a moving target in the radar monitoring scene according to the range-Doppler diagram.
Specifically, constant false alarm detection is carried out on the range Doppler graph, point clouds generated by moving targets are extracted, and the moving direction and the moving speed of the moving targets in the radar monitoring scene are determined according to the point clouds.
And S4, judging whether a moving target exists in the radar monitoring scene according to the multi-scale high-resolution range profile.
Specifically, calculating a first target existence feature and a second target existence feature of the multi-scale high-resolution range profile in adjacent frames, and judging whether a moving target exists in a radar monitoring scene according to the moving direction of the moving target, the first target existence feature and the second target existence feature;
s5, extracting characteristics of the multi-scale high-resolution range profile, the moving direction and the moving speed by using the characteristic extraction matrix, and judging the action type of the moving object at the current moment according to the category template and the corresponding action label, wherein the characteristic extraction matrix, the category template and the corresponding action label are dynamic data obtained by using the multi-scale high-resolution range profile, the moving direction and the moving speed in a period of time before the current moment.
In this embodiment, the feature extraction matrix, the category template and the behavior label are updated at regular time, so that the motion type of the moving object at the moment is determined to be judged directly according to the latest updated feature extraction matrix, category template and behavior label.
S6, performing health risk early warning on the moving target according to the action type of the moving target and the radar monitoring scene.
Specifically, the method for performing health risk early warning on the moving target according to the action type of the moving target and the radar monitoring scene comprises the following steps: counting action types of the moving object at a plurality of moments; and predicting the action types of the moving target at a plurality of moments according to the radar monitoring scene to obtain a health risk result.
For example: if the moving object is in a sitting posture for a long time, the user is indicated to have the risk of constipation, and the risk of dangerous actions such as falling and the like is increased; if the moving object uses the toilet more times and the using time is shorter each time, the risk that the user may have 'frequent urination' is indicated; if the moving object uses the toilet more times and each time the using time is longer, it is indicated that the user has a possibility of "pulling the stomach". If the moving object is in a 'static lying' posture in a bedroom for a long time, the user is possibly sleeping, and if the sleeping time is short, the user is possibly suffering from insomnia and insufficient sleeping, so that risks of 'poor energy' and 'absentmindedness' are increased; longer sleep time is associated with increased risk of dizziness, etc. The health risk result classification includes, but is not limited to, the above-mentioned several types, and may be reasonably set in combination with a specific application scenario.
According to the method, the radar echo signals received by radar equipment in a radar monitoring scene are preprocessed, moving target detection, behavior analysis and early warning are carried out, noise in the scene is reduced, accuracy of signal analysis is improved, information such as a multiscale high-resolution range profile, a range Doppler image and angles of moving targets relative to the radar can be obtained through preprocessing, detection of moving speeds and moving directions of the moving targets can be achieved, then whether the moving targets exist in the radar monitoring scene is judged, real-time tracking of the moving targets is achieved, human behavior action analysis is carried out on the moving targets in the radar monitoring scene by means of a feature extraction matrix, a category template and a behavior label which are updated regularly, targeted and self-adaptive action recognition can be achieved for differences of different testers, accurate health risk early warning is provided, occurrence risks of diseases, falling and other unexpected situations can be accurately predicted, and unexpected occurrence is reduced.
In addition, the millimeter wave radar can be used for identifying human behaviors and predicting health risks under the condition of not invading user privacy, can be applied to private scenes such as toilets, bedrooms and the like, and has a wider application range. And the basic information of the user can be analyzed by collecting the use condition of the user, the algorithm threshold is adaptively adjusted, the parameter setting which is too complex is not needed, the use is simpler and more convenient, the adaptability to different scenes and unnecessary people is higher, and the large-scale popularization is convenient.
The health risk early warning method based on the millimeter wave radar provided by the embodiment of the invention further comprises the following steps:
counting behavior labels of each type of moving targets in preset historical time;
and judging and analyzing the behavior habit of the moving target in the corresponding radar monitoring scene according to the counted behavior label.
Specifically, the actions within the history D1 days in step S6 are counted, and the behavior tag of the moving object at the current time is analyzed in combination with the installation position of the radar device, for example: when the device is installed in a bathroom, the sitting posture can represent the behaviors of a tester in using a toilet, the squatting posture can be used for washing clothes and the like, and the standing posture can be used for washing hands, showering and the like. In addition, the occurrence time of people reported by daily equipment can be counted, and the use frequency of the toilet can be analyzed; according to the movement speed of the testers in each toilet, the age, posture and other information of the users can be approximately analyzed; based on the posture and duration of the user, it is possible to infer whether the user is performing a washing, brushing, etc. action. The method is helpful for knowing the activity mode and the behavior rule of the target, so that references and bases are provided for subsequent behavior prediction, health evaluation and the like, misjudgment of the behaviors can be reduced, and the accuracy of the behavior prediction and evaluation is improved.
The feature extraction matrix, the category template and the corresponding behavior label in the step S5 are dynamic data obtained by utilizing a multiscale high-resolution range profile, a motion direction and a motion speed in a period of time before the current moment, and specifically include:
storing the multi-scale high-resolution range profile, the moving direction, the moving speed and the moving target judgment result to obtain a history reference characteristic; after radar echo signals are processed each time, the obtained data are stored to obtain historical reference data, wherein the multi-scale high-resolution range profile, the moving direction and the moving speed of a moving target are in one-to-one correspondence when the moving target exists and does not exist, the stored data are used as reference templates, and meanwhile, the stored historical data are subjected to coverage updating every D2 days, so that the data quantity is reduced and the data referential performance is improved.
Performing principal component analysis on the historical reference features at preset time intervals, and determining a principal component feature extraction matrix based on analysis results；
Matrix extraction using principal component featuresExtracting the characteristic of the history reference characteristic to obtain the main component characteristic +.>；
For principal component characteristicsClustering to obtain feature classification result, specifically, using the following formula to perform main component feature +. >Performing cluster analysis, and clustering data into C types:
，
wherein,for clustering, the class-C moving object features are obtained, and a cluster (level) represents a clustering algorithm, wherein the clustering algorithm comprises, but is not limited to, K-Means, DBSCAN and other clustering methods.
Respectively classifying various data in the results according to the characteristicsCorresponding multiscale high-resolution range profile, movement direction and movement speed determine various data +.>Corresponding behavior tags.
Further, principal component analysis is performed on the historical reference features, and a principal component feature extraction matrix is determined based on the analysis resultComprising:
the movement order exists in the history reference characteristics within the preset timeThe time-lapse multiscale high-resolution range profile, the motion direction and the motion speed are spliced to obtain characteristic data。
For characteristic data in slow time dimensionStandardized processing to obtain standard characteristic data +.>；
For standard characteristic dataPerforming principal component analysis, and determining principal component feature extraction matrix based on analysis result>。
Further, a principal component feature extraction matrix is determined based on the analysis resultComprising:
according to standard characteristic dataDetermining principal component feature extraction matrix by using feature value and feature vector obtained by principal component analysis>；
Specifically, principal component analysis was performed using the following formula:
，
Wherein,for the feature value corresponding to each dimension principal component, the dimension is +.>，/>Is a principal component analysis method.
Determining the dimension according to the ratio of the characteristic value of each dimension to the characteristic value of all dimensions；
Extracting matrix from principal component featuresExtracting maximum->Feature vectors corresponding to the dimensional feature values obtain a principal component feature extraction matrix +.>。
Further, the dimension is determined according to the ratio of the characteristic value of each dimension to the characteristic value of all the dimensionsComprising:
calculating the ratio of the sum of the eigenvalues of the previous i dimension to the sum of the eigenvalues of all dimensionsThe initial value of i is 1;
judging the ratioWhether or not threshold is reached +.>；
If the ratio isReaching threshold->Then the current i is taken as dimension +.>；
If the ratio isDoes not reach threshold +.>The value of i is increased.
Specifically, the obtained eigenvalues are ranked in order from high to low, and then the ratio of the sum of the eigenvalues of the previous i-dimension to the sum of the eigenvalues of all dimensions is calculated from the ranking by the following formula：
，
Wherein,is the eigenvalue of the j-th dimension.
Then calculating the ratio of the sum of the eigenvalues of the previous i dimension to the sum of the eigenvalues of all dimensionsThen, is matched with the threshold valueComparing, if the characteristic value duty ratio reaches the threshold value +.>The corresponding dimension i is taken as dimension +. >If the characteristic value is not up to the threshold +.>Continue to increase iIs calculated as the characteristic value of +.>Until threshold +.>Stopping calculation, and then extracting the principal component feature extraction matrix +.>Middle and maximum->Feature vector corresponding to the dimension feature value forms a principal component feature extraction matrix +.>The method comprises the steps of carrying out a first treatment on the surface of the Thus, in the principal component feature extraction matrix +.>In each updating, main component analysis is carried out once according to the continuously updated historical reference characteristics, and a new main component characteristic extraction matrix is obtained according to the analysis result>。
According to the method, the device and the system, the main component analysis is carried out on the characteristics of the moving target in the stored historical reference characteristics at intervals, so that the most representative characteristics of the moving target in the latest data can be extracted, and the accuracy of the data is improved; by the method, the actual data of various scenes can be adaptively updated, so that the applicability of an algorithm is improved; then, by grouping the updated data according to similarity, grouping the similar data together to form different categories, moving objects can be classified according to their characteristics; finally, the height of each type of moving object is calculated, the behavior label of each type of feature can be determined according to the difference of the heights, and the behavior of the current moving object can be determined according to the reference data by continuously updating the reference data, so that accurate risk early warning can be conveniently carried out.
Further, various data in the results are classified according to the characteristics respectivelyCorresponding multi-scale high-resolution range profile, motion direction, motion speed, radar height and angle of moving target relative to radar to determine various dataA corresponding behavior tag comprising:
extracting the target height at the corresponding moment according to the multi-scale high-resolution range profile, the radar height and the angle;
calculating various dataAverage height of (2);
determining various data according to average heightObtaining a category template;
and determining the behavior label under each cluster label according to the movement direction and movement speed corresponding to each cluster label.
Specifically, for various types of dataThe corresponding multi-scale high-resolution range profile, the motion direction, the motion speed and the energy value are analyzed, namely, for each type of moving target, a plurality of distances with the maximum energy value in the fine-scale high-resolution range profile are extracted, meanwhile, the historical first moving target distances extracted by the adjacent historical moment aiming at the corresponding fine-scale high-resolution range profile are obtained, the difference value between the extracted distances and the historical first moving target distances is calculated respectively, and then the smallest distance in the difference value is selected as the first moving target distance at the corresponding moment of the type of moving target.
If the moving speed of the moving object is zero, extracting a preset number of distances from the coarse-scale high-resolution range profile, respectively calculating the difference value between the extracted distances and the second moving object distances at the adjacent historical moment, selecting the distance with the smallest difference value as the second moving object distance of the moving object at the corresponding moment, and if the moving speed of the moving object is not zero, determining the first moving object distance as the second moving object distance of the moving object at the corresponding moment.
Calculating the height of the moving object according to the angle of the moving object relative to the radar, the distance of the second moving object and the radar height;
as shown in fig. 1, the height of the moving object is calculated from the angle of the moving object with respect to the radar in the vertical direction and the horizontal direction extracted in step S2, the height of the radar from the ground, and the calculation formula is as follows:
，
wherein,representing the extracted second moving object distance, < > and>is of circumference rate>Indicating that the distance in the vertical direction extracted in step S2 is +.>Angle of the moving object relative to the radar, < +.>Indicating that the distance in the horizontal direction extracted in step S2 is +.>Angle of the moving object relative to the radar, < +.>Representing the radar height from the ground, +. >Representing the height of the moving object from the ground, which is the head-to-ground distance.
Then calculating the average height, the moving direction and the moving speed of the moving targets, defining the specific behaviors corresponding to each type of moving targets, and defining the category templates (clustering labels) of each type of moving targets as follows according to the height threshold value: standing, sitting, bending over, squatting and lying 5 kinds. The behavior is then further subdivided according to the speed and direction of motion of the moving object for each class of behavior tags, for example: if the height corresponding to a certain type of moving object is a standing posture, and meanwhile, the moving speed of the moving object is high, and the moving direction is a direction close to a radar, the behavior label of the moving object can be specifically defined as 'entering a room rapidly'; if the height corresponding to a certain type of moving object is a standing posture, and meanwhile, the moving speed of the moving object is high, and the moving direction is far away from the radar direction, the behavior label of the moving object can be specifically defined as "leave a room rapidly"; if the height corresponding to a certain type of moving object is a standing posture, and meanwhile, the moving speed of the moving object is moderate, and the moving direction is a direction close to a radar, the behavior label of the moving object can be specifically defined as entering a room; if the height corresponding to a certain type of moving object is a standing posture, and meanwhile, the moving speed of the moving object is moderate and the moving direction is far away from the radar direction, the behavior label of the moving object can be specifically defined as leaving a room; if the height corresponding to a certain type of moving object is a standing posture, and meanwhile, the moving speed of the moving object is slower and the moving direction is changed faster, the behavior label of the moving object can be specifically defined as standing movement; if the height corresponding to a certain type of moving object is a standing posture, and meanwhile, the moving speed of the moving object is lower and the moving energy value is lower, the behavior label of the moving object can be specifically defined as standing; if the height corresponding to a certain type of moving object is sitting, and the moving energy value of the moving object is lower, the behavior label of the moving object can be specifically defined as "sitting"; if the height corresponding to a certain type of moving object is a sitting posture and the moving energy value of the moving object is higher, the behavior label of the moving object can be specifically defined as a 'moving sitting posture'; if the corresponding height of a certain type of moving object is squatting, and the moving energy value of the moving object is lower, the behavior label of the moving object can be specifically defined as 'static squat'; if the corresponding height of a certain type of moving object is squatting, and the moving energy value of the moving object is higher, the behavior label of the moving object can be specifically defined as 'moving squatting'; if the corresponding height of a certain type of moving object is lying, and the moving energy value of the moving object is low, the behavior label of the moving object can be specifically defined as lying; if the corresponding height of a certain type of moving object is a lying posture and the moving energy value of the moving object is higher, the behavior label of the moving object can be specifically defined as a 'moving lying posture', and the like; the behavior definition includes, but is not limited to, the above-mentioned example human behaviors, and can be properly defined according to actual requirements.
In this embodiment, the category templates and corresponding behavior labels will be extracted along with the principal component features to obtain a matrixUpdating, namely, extracting the distance corresponding to the strongest energy point in the fine-scale high-resolution range profile according to the category of the moving object, calculating the difference value between the distance and the historical first moving object, determining the first moving object distance of each category of moving object at the moment, and improving the accuracy of the moving object distance by selecting the distance with the smallest difference value; the second moving object distance at that moment is determined according to the moving speed or the moving direction of the moving object, and the height of each moving object can be calculated according to the angle of the moving object relative to the radar, the second moving object distance and the radar height, so that information is provided for three-dimensional positioning of the moving object, accurate positioning and behavior analysis of the moving object are realized, behaviors of the moving object are analyzed, a behavior label of the moving object is formed, and reference is made for later behavior analysis.
Step S5 further comprises:
multiscale high resolution when moving objects are to be presentThe image separation, the movement direction and the movement speed are spliced to obtain real-time characteristic data；
For real-time feature data in the slow time dimension Performing normalization processing to obtain real-time standard characteristic data +.>；
Matrix extraction using principal component featuresFor real-time standard feature data->Extracting features to obtain real-time main component features ∈ ->。
In the embodiment, the multi-scale high-resolution range profile, the motion direction and the motion speed when the moving target exists are spliced and standardized to obtain real-time standard characteristic dataAnd principal component feature extraction matrix obtained from historical reference features +.>Extracting principal component to obtain current principal component characteristic ∈>Furthermore, the main component is characterized according to the category template>Classifying, and matching with behavior label to obtain motion type of moving object, wherein the principal component extraction matrix, class template and behavior label are updated continuouslyAnd the accuracy of the analysis result is improved.
In a preferred embodiment, S2, preprocessing a radar echo signal to obtain a multi-scale high-resolution range profile, a range doppler plot, and an angle of a moving target relative to a radar, includes:
s21, extracting a distance dimension complex signal of a radar echo signal, and extracting a moving target from the distance dimension complex signal according to a time scale to obtain a fine-scale high-resolution range profile and a coarse-scale high-resolution range profile;
Specifically, processing such as direct current removal, windowing, fast fourier transformation and the like can be performed on each Chirp signal in the radar echo signals respectively to obtain a first distance dimension complex signal; carrying out slow time DC removal processing on N first distance dimension complex signals in each frame to obtain second distance dimension complex signals; and averaging N first distance dimension complex signals in each frame in the time dimension to obtain a third distance dimension complex signal. And (3) calculating absolute values of the second distance dimension complex signals in each frame, and calculating an average value in a slow time dimension to obtain a fine-scale high-resolution range profile. And (3) taking the third distance dimension complex signals obtained by the adjacent K frames to respectively perform DC removal, absolute value calculation and average value calculation in the slow time dimension, so as to obtain a coarse-scale high-resolution range profile. The operations of DC removal, windowing, fast Fourier transformation, absolute value calculation and mean value calculation are well known and accepted public technologies in the field.
S22, processing the distance dimension complex signal in a slow time dimension, and extracting a distance Doppler graph of the moving target;
specifically, windowing, fast Fourier transforming and absolute value taking processing are carried out on the second distance dimension complex signal in a slow time dimension, and then moving target extraction is carried out, so that a distance Doppler graph is obtained.
S23, performing conjugate multiplication on the distance dimension complex signals extracted by different radar antennas, and obtaining an average value to obtain phase-dependent signals in different directions, and extracting angles of moving targets in different directions relative to the radar according to the phase-dependent signals in different directions.
As shown in fig. 3, the millimeter wave radar device Tx includes three receiving antennas arranged in different directions, respectively, connected to each otherExtracting a first distance dimension complex signal, a second distance dimension complex signal and a third distance dimension complex signal from the received signal, wherein the center distance between the antenna Rx1 and the antenna Rx3 in the vertical direction isThe center-to-center distances of the antennas Rx2 and Rx3 in the horizontal direction are +.>. Conjugate multiplying the second distance dimension complex signal received by the antenna Rx1 and the second distance dimension complex signal received by the antenna Rx3 to obtain a first coherent signal in the vertical direction; the second distance-dimensional complex signal received by the antenna Rx2 is conjugate-multiplied with the second distance-dimensional complex signal received by the antenna Rx3 to obtain a first correlation signal in the horizontal direction. Averaging the first phase-related signals in the vertical direction in a slow time dimension to obtain second phase-related signals in the vertical direction; averaging the first phase-reference signals in the horizontal direction in a slow time dimension to obtain second phase-reference signals in the horizontal direction, and then respectively calculating phase angles of the second phase-reference signals in the vertical direction and the second phase-reference signals in the horizontal direction by using the following formula, and extracting to obtain angles of moving targets in the vertical direction and the horizontal direction relative to the radar:
，
Wherein,is the imaginary part of the second correlation signal on the distance gate of r>Is the real part of the second correlation signal on the distance gate No. r,/>For the angle of the moving object relative to the radar on the distance gate No. r +.>Indicates the direction, and->13, vertical direction, ">23, horizontal direction,/->Is an arctangent function, a well-known and publicly known technique in the art.
The fine-scale high-resolution range profile extracted in the embodiment can provide more detailed limb movement information, and the coarse-scale high-resolution range profile can provide the overall movement condition of a moving target; the extracted range-Doppler image can provide information of the moving target in the dimensions of range, speed and the like so as to accurately analyze the motion state of the moving target; and then, the distance dimension complex signals extracted by the radar antennas in different directions are utilized to obtain phase-related signals in different directions, and the angle information of the moving target relative to the radar can be extracted by analyzing the phase-related signals in different directions, so that the azimuth angle of the moving target in the horizontal and vertical directions can be determined, and the moving target can be positioned and tracked conveniently.
In a preferred embodiment, S3, determining a moving direction and a moving speed of a moving target in a radar monitoring scene according to a range-doppler plot includes:
S31, performing constant false alarm detection on the range Doppler graph to obtain a detection threshold;
in this embodiment, an ordered statistics-type constant false alarm detector (OS-CFAR) is used to process the range-doppler plot in a sliding window manner, signals in the windows are ordered in each window, and then a proper ordering position is selected as a threshold, i.e. a detection threshold, according to a preset false alarm probability (i.e. a desired false alarm rate).
S32, dividing the range-Doppler graph into a radar area, a radar area and an in-situ stay area according to Doppler values, and respectively calculating energy sums of point clouds higher than a detection threshold in each area to obtain three Doppler energy sums;
in this embodiment, the detection threshold includes a first doppler threshold and a second doppler threshold, the first doppler threshold is greater than zero, the second doppler value is less than zero, the distance doppler map is divided into three areas according to the doppler values of the point clouds by traversing the point clouds in the sliding window, and then the energy sum of the point clouds in the three areas is calculated, where the energy is the amplitude of the signal, specifically: dividing a point cloud with a Doppler value smaller than a first Doppler threshold value into a radar approaching area, and calculating the energy value of the point cloud in the radar approaching area to obtain a first Doppler energy sum; dividing the point cloud with the Doppler value larger than the second Doppler threshold value into a far-away radar area, and calculating the energy sum of the point cloud in the far-away radar area to obtain a second Doppler energy sum; dividing the point cloud with the Doppler value larger than the first Doppler threshold value and smaller than the second Doppler threshold value into in-situ stay areas, and calculating the energy sum of the point cloud in the in-situ stay areas to obtain a third Doppler energy sum.
S33, comparing the three Doppler energy sums with three motion direction discrimination thresholds to determine the motion direction of the moving target;
in this embodiment, since the total signal numbers of different areas may be different, the magnitude of the energy sum may not be comparable, and secondly, the energy sum may be affected by the noise signal, so that the comparison result is inaccurate, therefore, by calculating the difference between the first doppler energy sum and the second doppler energy sum, and comparing the difference with the motion direction discrimination threshold, a more reliable determination result is obtained, which specifically is: if the difference value is smaller than the first motion direction judging threshold value, the motion direction of the moving object is determined to be close to the radar, if the difference value is larger than the second motion direction judging threshold value, the motion direction of the moving object is determined to be far away from the radar, and if the third Doppler energy sum is larger than the third motion direction judging threshold value, the motion direction of the moving object is determined to be in-situ stop.
S34, determining the moving speed of the moving object according to the Doppler value of the area corresponding to the moving direction of the moving object.
In this embodiment, the moving speed of the moving object is determined according to the moving direction of the moving object determined in step S33 and according to the doppler value of the area where the strongest energy of the moving direction is located.
In the embodiment, first, constant false alarm detection is performed on a range-doppler plot, and a proper threshold value (doppler value) is set to screen out a region which is possibly a moving target and exclude false alarm noise; dividing a range Doppler diagram into a radar approaching region, a radar far region and a radar in-situ stay region according to a detection threshold, respectively calculating energy sums of point clouds in the three regions to represent the intensity or energy distribution condition of moving targets in different regions, comparing the three Doppler energy sums with corresponding movement direction discrimination thresholds to determine the movement direction of the moving targets, and determining whether the moving targets are stationary, approaching to the radar or moving away from the radar according to the size relation between the energy sums and the thresholds; and finally, determining the moving speed of the moving object according to the Doppler value in the area corresponding to the moving direction of the moving object, and providing a basis for subsequent moving object analysis and early warning.
In a preferred embodiment, S4, determining whether a moving target exists in the radar monitoring scene according to the multi-scale high-resolution range profile includes:
s41, calculating a first target existence feature according to probability distribution of energy values of the coarse-scale high-resolution range profile;
Specifically, statistics is carried out on energy data corresponding to a coarse-scale high-resolution range profile of an adjacent P1 frame, a histogram is established, then the frequency of each interval in the histogram is normalized, the probability value of each interval is calculated, and the probability distribution of the energy value of the coarse-scale high-resolution range profile is obtainedCalculating probability distribution->Is then characterized by the presence of +.>。/>
S42, determining the existence characteristic of the second target according to the relativity of the energy value and the energy mean value of the coarse-scale high-resolution range profile;
specifically, the energy data of the coarse-scale high-resolution range profile of the adjacent P2 frame is counted, and the energy mean value is calculated as a templateThen calculating the energy value and the template of the coarse-scale high-resolution range profile at the current moment>Is the correlation of the second object, phase Guan Du as the second object presence feature +.>。
S43, if the first target existence characteristic is larger than the first unmanned detection threshold value and the second target existence characteristic is larger than the second unmanned detection threshold value, judging that a moving target exists in the radar monitoring scene at the current moment.
In this embodiment, since the energy dispersion may reflect the distribution of the moving objects in the image, and the correlation degree may reflect the correlation degree of the moving objects in the image, by setting an appropriate threshold, it may be determined whether a person is in the scene at the current time, and features exist in the first object Greater than a first threshold for detection of presence of a person and a second target presence feature +.>And when the detection threshold value is larger than the second unmanned detection threshold value, judging that a person exists in the radar monitoring scene at the current moment, wherein the first unmanned detection threshold value is an energy dispersion threshold value, and the second unmanned detection threshold value is a correlation threshold value.
In the embodiment, the discrete degree of the energy value is estimated by calculating the energy dispersion (variance or standard deviation) of the probability distribution, so that the moving target can be subjected to preliminary feature extraction and analysis on a coarse scale; the correlation degree between the energy value at the current moment and the energy mean value of the past frame can be estimated through calculating the correlation degree, and the feature extraction and analysis are carried out on the moving target on the time sequence; in a word, whether a moving target exists at the current moment or not is judged by calculating energy dispersion and correlation degree and comparing the energy dispersion and correlation degree with a preset threshold value, so that the accuracy and the reliability of moving target detection and scene analysis are improved.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It is apparent that the above examples are given by way of illustration only and are not limiting of the embodiments. Other variations or modifications of the above teachings will be apparent to those of ordinary skill in the art. It is not necessary here nor is it exhaustive of all embodiments. And obvious variations or modifications thereof are contemplated as falling within the scope of the present invention.
Claims (11)
1. A health risk early warning method based on millimeter wave radar is characterized by comprising the following steps:
collecting radar echo signals in a radar monitoring scene;
preprocessing the radar echo signals to obtain a multi-scale high-resolution range profile, a range Doppler image and angles of moving targets relative to the radar;
Determining the motion direction and the motion speed of a moving target in a radar monitoring scene according to the range-Doppler diagram;
judging whether a moving target exists in the radar monitoring scene according to the multi-scale high-resolution range profile;
extracting characteristics of the multi-scale high-resolution range profile, the motion direction and the motion speed by using a characteristic extraction matrix, and judging the action type of the moving object at the current moment according to a category template and a corresponding action label, wherein the characteristic extraction matrix, the category template and the corresponding action label are dynamic data obtained by using the multi-scale high-resolution range profile, the motion direction and the motion speed in a period of time before the current momentThe method specifically comprises the following steps: storing the multi-scale high-resolution range profile, the moving direction, the moving speed and the moving target judgment result to obtain a history reference characteristic; performing principal component analysis on the historical reference features at preset time intervals, and determining a principal component feature extraction matrix based on analysis resultsThe method comprises the steps of carrying out a first treatment on the surface of the Extracting matrix by using the principal component features>Extracting the characteristic of the history reference characteristic to obtain main component characteristic +.>The method comprises the steps of carrying out a first treatment on the surface of the For the main component feature- >Clustering is carried out to obtain a feature classification result; extracting the target height at the corresponding moment according to the multi-scale high-resolution range profile, the radar height and the angle; calculating various data in the feature classification result>Average height of (2); determining various data according to the average height>Obtaining a category template; determining behavior labels under each cluster label according to the movement direction and movement speed corresponding to each cluster label;
and carrying out health risk early warning on the moving target according to the action type of the moving target and the radar monitoring scene.
2. The method of claim 1, wherein the historical reference features are subjected to principal component analysis, and a principal component feature extraction matrix is determined based on the analysis resultsComprising:
splicing the multi-scale high-resolution range profile, the motion direction and the motion speed when the moving target exists in the historical reference characteristics within the preset time to obtain characteristic data；
For the characteristic data in the slow time dimensionStandardized processing to obtain standard characteristic data +.>；
For the standard characteristic dataPerforming principal component analysis, and determining principal component feature extraction matrix based on analysis result >。
3. The method of claim 2, wherein the principal component feature extraction matrix is determined based on the analysis resultComprising:
according to the standard characteristic dataDetermining principal component feature extraction matrix by using feature value and feature vector obtained by principal component analysis>；
According to the characteristic value of each dimension accounting for the characteristics of all dimensionsRatio of values to determine dimension；
Extracting matrix from the principal component featuresExtracting maximum->Feature vectors corresponding to the dimensional feature values obtain a principal component feature extraction matrix +.>。
4. A method according to claim 3, wherein dimensions are determined from the ratio of the feature values of each dimension to the feature values of all dimensionsComprising:
calculating the ratio of the sum of the eigenvalues of the previous i dimension to the sum of the eigenvalues of all dimensionsThe initial value of i is 1;
judging the ratioWhether or not threshold is reached +.>；
If the ratio isReaching threshold->Then the current i is taken as dimension +.>；
If the ratio isDoes not reach threshold +.>The value of i is increased.
5. The method of claim 1, wherein the feature extraction of the multi-scale high resolution range profile, direction of motion, speed of motion using a feature extraction matrix comprises:
splicing the multiscale high-resolution range profile, the motion direction and the motion speed when the moving target exists to obtain real-time characteristic data ；
For the real-time feature data in the slow time dimensionPerforming normalization processing to obtain real-time standard characteristic data +.>；
Extracting matrix by using the principal component featuresFor the real-time standard characteristic data +.>Extracting features to obtain real-time main component features ∈ ->。
6. The method of claim 1, wherein preprocessing the radar echo signals to obtain a multi-scale high-resolution range profile, a range-doppler profile, and an angle of a moving object relative to a radar, comprises:
extracting a distance dimension complex signal of the radar echo signal, and extracting a moving target from the distance dimension complex signal according to a time scale to obtain a fine-scale high-resolution range profile and a coarse-scale high-resolution range profile;
processing the distance dimension complex signal in a slow time dimension, and extracting a distance Doppler graph of a moving target;
and carrying out conjugate multiplication on the distance dimension complex signals extracted by different radar antennas, and solving an average value to obtain phase-coherent signals in different directions, and extracting angles of moving targets in different directions relative to the radar according to the phase-coherent signals in different directions.
7. The method of claim 1, wherein determining a direction of motion and a speed of motion of a moving object within a radar surveillance scene from the range-doppler plot comprises:
Performing constant false alarm detection on the range-Doppler graph to obtain a detection threshold;
dividing the range-Doppler graph into a radar area, a radar area and an in-situ stay area according to Doppler values, and respectively calculating energy sums of point clouds in each area, which are higher than a detection threshold, so as to obtain three Doppler energy sums;
comparing the three Doppler energy sums with three motion direction discrimination thresholds to determine the motion direction of the moving target;
and determining the moving speed of the moving object according to the Doppler value of the area corresponding to the moving direction of the moving object.
8. The method of claim 1, wherein determining whether a moving object is present in a radar surveillance scene based on the multi-scale high-resolution range profile comprises:
calculating the first target existence characteristic according to the probability distribution of the energy value of the coarse-scale high-resolution range profile;
determining the existence characteristic of the second target according to the relativity of the energy value and the energy mean value of the coarse-scale high-resolution range profile;
if the first target existence characteristic is larger than the first unmanned detection threshold value and the second target existence characteristic is larger than the second unmanned detection threshold value, determining that a moving target exists in the radar monitoring scene at the current moment.
9. The method of claim 1, wherein performing health risk pre-warning on the moving object according to the action type of the moving object and the radar monitoring scene comprises:
counting action types of the moving object at a plurality of moments;
and predicting the action types of the moving target at a plurality of moments according to the radar monitoring scene to obtain a health risk result.
10. The method as recited in claim 1, further comprising:
counting behavior labels of each type of moving targets in preset historical time;
and judging the behavior habit of each type of moving object in the corresponding radar monitoring scene according to the statistical behavior label.
11. Health risk early warning equipment based on millimeter wave radar, characterized by comprising: a processor and a memory coupled to the processor; wherein the memory stores instructions executable by the processor to cause the processor to perform the millimeter wave radar-based health risk warning method of any one of claims 1-10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577651.2A CN117281498B (en) | 2023-11-24 | 2023-11-24 | Health risk early warning method and equipment based on millimeter wave radar |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311577651.2A CN117281498B (en) | 2023-11-24 | 2023-11-24 | Health risk early warning method and equipment based on millimeter wave radar |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117281498A CN117281498A (en) | 2023-12-26 |
| CN117281498B true CN117281498B (en) | 2024-02-20 |
Family
ID=89248385
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311577651.2A Active CN117281498B (en) | 2023-11-24 | 2023-11-24 | Health risk early warning method and equipment based on millimeter wave radar |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117281498B (en) |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107920737A (en) * | 2015-05-31 | 2018-04-17 | Sens4保护公司 | The remote supervision system of mankind's activity |
| CN111680539A (en) * | 2020-04-14 | 2020-09-18 | 北京清雷科技有限公司 | Dynamic gesture radar recognition method and device |
| CN112837504A (en) * | 2021-02-09 | 2021-05-25 | 中云沃达（山东）物联网科技有限公司 | Nursing device, system and monitored control system based on millimeter wave detection |
| CN113397520A (en) * | 2021-07-14 | 2021-09-17 | 北京清雷科技有限公司 | Information detection method and device for indoor object, storage medium and processor |
| CN114442079A (en) * | 2022-01-14 | 2022-05-06 | 北京清雷科技有限公司 | Target object falling detection method and device |
| CN114895301A (en) * | 2022-05-23 | 2022-08-12 | 武汉大学 | Millimeter wave radar and video-assisted indoor fall detection method and device |
| CN115345908A (en) * | 2022-10-18 | 2022-11-15 | 四川启睿克科技有限公司 | Human body posture recognition method based on millimeter wave radar |
| CN115356725A (en) * | 2022-08-27 | 2022-11-18 | 苏州智聚芯联微电子有限公司 | Integrated non-contact millimeter wave radar system for health monitoring |
| CN115372963A (en) * | 2022-10-24 | 2022-11-22 | 北京清雷科技有限公司 | Fall-down behavior multi-level detection method and device based on millimeter wave radar signals |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014055755A1 (en) * | 2012-10-05 | 2014-04-10 | TransRobotics, Inc. | Systems and methods for high resolution distance sensing and applications |
| CA2891839C (en) * | 2014-05-16 | 2023-02-14 | Mohamed Mabrouk | Remote sensing of human breathing at a distance |
| WO2022058735A2 (en) * | 2020-09-16 | 2022-03-24 | Nodens Medical Ltd | Millimeterwave radar system for determining an activity record |
| CN112505648B (en) * | 2020-11-19 | 2023-06-30 | 西安电子科技大学 | Target feature extraction method based on millimeter wave radar echo |
| TWI804072B (en) * | 2021-11-26 | 2023-06-01 | 國立陽明交通大學 | Real-time object motion state recognition method using millimeter wave radar |
-
2023
- 2023-11-24 CN CN202311577651.2A patent/CN117281498B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107920737A (en) * | 2015-05-31 | 2018-04-17 | Sens4保护公司 | The remote supervision system of mankind's activity |
| CN111680539A (en) * | 2020-04-14 | 2020-09-18 | 北京清雷科技有限公司 | Dynamic gesture radar recognition method and device |
| CN112837504A (en) * | 2021-02-09 | 2021-05-25 | 中云沃达（山东）物联网科技有限公司 | Nursing device, system and monitored control system based on millimeter wave detection |
| CN113397520A (en) * | 2021-07-14 | 2021-09-17 | 北京清雷科技有限公司 | Information detection method and device for indoor object, storage medium and processor |
| CN114442079A (en) * | 2022-01-14 | 2022-05-06 | 北京清雷科技有限公司 | Target object falling detection method and device |
| CN114895301A (en) * | 2022-05-23 | 2022-08-12 | 武汉大学 | Millimeter wave radar and video-assisted indoor fall detection method and device |
| CN115356725A (en) * | 2022-08-27 | 2022-11-18 | 苏州智聚芯联微电子有限公司 | Integrated non-contact millimeter wave radar system for health monitoring |
| CN115345908A (en) * | 2022-10-18 | 2022-11-15 | 四川启睿克科技有限公司 | Human body posture recognition method based on millimeter wave radar |
| CN115372963A (en) * | 2022-10-24 | 2022-11-22 | 北京清雷科技有限公司 | Fall-down behavior multi-level detection method and device based on millimeter wave radar signals |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117281498A (en) | 2023-12-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2021004299A1 (en) | Human detection method and device, electronic apparatus, and storage medium | |
| JP5301973B2 (en) | Crime prevention device and program | |
| US20210116540A1 (en) | Integration of Tracking with Classifier in mmWave Radar | |
| CN110456320B (en) | Ultra-wideband radar identity recognition method based on free space gait time sequence characteristics | |
| CN114942434B (en) | Fall gesture recognition method and system based on millimeter wave Lei Dadian cloud | |
| CN108171193B (en) | Polarized SAR (synthetic aperture radar) ship target detection method based on super-pixel local information measurement | |
| CN114814832A (en) | Millimeter wave radar-based real-time monitoring system and method for human body falling behavior | |
| CN111507361B (en) | Action recognition device, method and system based on microwave radar | |
| CN116338684A (en) | Human body falling detection method and system based on millimeter wave radar and deep learning | |
| CN115469301A (en) | Old people activity capability assessment method and system based on ultra-wideband radar | |
| CN114442079A (en) | Target object falling detection method and device | |
| CN117281498B (en) | Health risk early warning method and equipment based on millimeter wave radar | |
| CN115372963B (en) | Fall-down behavior multi-level detection method and equipment based on millimeter wave radar signals | |
| CN114140997A (en) | System and method for monitoring and early warning of retention and physical condition of old people in toilet of nursing home | |
| Ha et al. | Detection and localization of multiple human targets based on respiration measured by IR-UWB radars | |
| Xue et al. | A machine learning-based positioning method for poultry in cage environments | |
| AU2020417054A1 (en) | State detection | |
| CN116520315A (en) | Target recognition system, target recognition method and target recognition device | |
| CN115982620A (en) | Millimeter wave radar human body falling behavior identification method and system based on multi-class three-dimensional features and Transformer | |
| TWI809241B (en) | Life form radar system, identification method, and database establishment method | |
| CN111025288B (en) | Security radar monitoring device and system | |
| CN114510967A (en) | Interference signal filtering method and device, electronic equipment and storage medium | |
| Lakshmanan et al. | A spatiotemporal approach to tornado prediction | |
| Fernandez et al. | Cane with millimeter wave radar for base of support measurement | |
| CN117315886B (en) | UWB radar-based method and device for detecting impending falling of personnel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |