CN114236492B - Millimeter wave radar micro gesture recognition method - Google Patents
Millimeter wave radar micro gesture recognition method Download PDFInfo
- Publication number
- CN114236492B CN114236492B CN202210164177.XA CN202210164177A CN114236492B CN 114236492 B CN114236492 B CN 114236492B CN 202210164177 A CN202210164177 A CN 202210164177A CN 114236492 B CN114236492 B CN 114236492B
- Authority
- CN
- China
- Prior art keywords
- signal
- difference frequency
- sample
- recognition
- echo
- 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 24
- 238000012549 training Methods 0.000 claims abstract description 40
- 238000012216 screening Methods 0.000 claims abstract description 9
- 238000012545 processing Methods 0.000 claims description 14
- 238000012360 testing method Methods 0.000 claims description 14
- 238000012544 monitoring process Methods 0.000 claims description 11
- 238000012163 sequencing technique Methods 0.000 claims description 11
- 238000010586 diagram Methods 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 4
- 238000012986 modification Methods 0.000 claims description 4
- 230000004048 modification Effects 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 3
- 230000003595 spectral effect Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 4
- 230000005540 biological transmission Effects 0.000 abstract 1
- 239000011159 matrix material Substances 0.000 description 24
- 230000009471 action Effects 0.000 description 13
- 238000012937 correction Methods 0.000 description 9
- 239000013598 vector Substances 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 238000010606 normalization Methods 0.000 description 4
- 238000005457 optimization Methods 0.000 description 4
- 238000005259 measurement Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000001228 spectrum Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
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
- 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
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Remote Sensing (AREA)
- Radar, Positioning & Navigation (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Computer Networks & Wireless Communication (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Medical Informatics (AREA)
- Evolutionary Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Electromagnetism (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a millimeter wave radar micro-motion gesture recognition method, which belongs to the technical field of gesture recognition and comprises the steps of transmitting a linear frequency modulation signal through a millimeter wave radar, receiving an echo signal reflected by a hand of a human body through a signal receiver, and acquiring a difference frequency combination set according to a transmission signal set formed by the linear frequency modulation signal and an echo signal set formed by the echo signal; calculating and obtaining the matching degree of the difference frequency combination set and the difference frequency sample set, comparing and screening the matching range formed by the matching degree and the difference frequency sample set, and setting the echo signal set meeting the standard as a selected signal set; the difference frequency sample set is a set which is obtained by training sample gestures and contains a plurality of sample difference frequency signals; the matching degree is used for analyzing and judging whether the difference frequency signal meets the requirement of gesture recognition; the method and the device are used for solving the technical problem that invalid data are not processed in advance when gesture recognition is carried out through a millimeter wave radar in the existing scheme, so that the recognition effect is poor.
Description
Technical Field
The invention relates to the technical field of gesture recognition, in particular to a millimeter wave radar micro gesture recognition method.
Background
The millimeter wave radar works in millimeter wave band, and the wavelength of the millimeter wave is between centimeter wave and light wave, so the millimeter wave has the advantages of microwave guidance and photoelectric guidance.
The existing gesture recognition scheme based on the millimeter wave radar generally processes an acquired echo signal, then inputs characteristic parameters in the echo signal into a constructed training model for training and outputs a recognition result, wherein the recognition result comprises recognition success and recognition failure.
However, the existing gesture recognition scheme has the defects that the collected echo signals are not analyzed and judged so as to determine whether the gesture generated by the human hand is in the recognition range of the recognition model, the echo signals which are not in the recognition range are trained and recognized, the overall gesture recognition effect is affected by the processing of invalid data, and the recognition model is not optimized and updated based on the recognition result.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a millimeter wave radar micro-motion gesture recognition method, which solves the following technical problems: how to solve and to have not handled the technical problem that invalid data leads to the recognition effect poor in advance when carrying out gesture recognition through the millimeter wave radar among the current scheme.
The purpose of the invention can be realized by the following technical scheme:
a millimeter wave radar micro gesture recognition method comprises the following specific steps:
s1: constructing a recognition database of sample gestures, wherein the recognition database comprises a first sample set and a second sample set of different states of the hand of the human body; the first sample set comprises various data of a mobile test of a human hand in a vertical state, and specifically comprises a first reference signal and a first training signal set; the second sample set comprises various data of a mobile test of a human hand in a horizontal state, and specifically comprises a second reference signal and a second training signal set;
s2: performing difference frequency processing and ADC (analog to digital converter) sampling on the test signal and the first sample set and the second sample set respectively to obtain a difference frequency sample set containing sample difference frequency signals;
s3: acquiring two-dimensional distance-Doppler image sets corresponding to a first sample set and a second sample set, and respectively processing and dividing to obtain a first division set corresponding to the first sample set and a second division set corresponding to the second sample set;
s4: transmitting a linear frequency modulation signal through a millimeter wave radar, receiving an echo signal reflected by a hand of a human body through a signal receiver, and acquiring a difference frequency combination set according to a transmitting signal set formed by the linear frequency modulation signal and an echo signal set formed by the echo signal;
s5: calculating and obtaining the matching degree of the difference frequency combination set and the difference frequency sample set, comparing and screening the matching range formed by the matching degree and the difference frequency sample set, and setting the echo signal set meeting the standard as a selected signal set; the difference frequency sample set is a set which is obtained by training sample gestures and contains a plurality of sample difference frequency signals; the matching degree is used for analyzing and judging whether the difference frequency signal meets the requirement of gesture recognition;
s6: filtering and monitoring echo signals in the selected signal set to obtain an echo monitoring set, acquiring a two-dimensional distance-Doppler image set corresponding to the echo monitoring set and setting the two-dimensional distance-Doppler image set as an identification image set, acquiring characteristic peak point coordinates of a trace map corresponding to the identification image set and sequencing the characteristic peak point coordinates according to time to obtain an identification sequencing set; dividing the identification sorting set according to a preset division number to obtain an identification division set;
s7: and matching and recognizing the recognition sequencing set and a recognition database of the pre-constructed sample gestures to obtain a recognition result of the human body gestures, and optimizing and updating the pre-constructed recognition database according to the recognition result.
Further, when the human hand is in a vertical state, setting the position of the human hand when the human hand is not moved as a first reference position, arranging and combining different positions of the human hand after moving according to time to obtain a first training position set, transmitting a test signal through a millimeter wave radar, receiving an echo signal reflected by the human hand through a signal receiver, and setting the echo signal reflected by the first reference position as a first reference signal; combining a plurality of echo signals transmitted by a first training position set according to time arrangement to obtain a first training signal set; the first reference signal and the first set of training signals constitute a first set of samples.
Further, when the human hand is in a horizontal state, setting the position of the human hand when the human hand is not moved as a second reference position, arranging and combining different positions of the human hand after moving according to time to obtain a second training position set, transmitting a test signal through a millimeter wave radar, receiving an echo signal reflected by the human hand through a signal receiver, and setting the echo signal reflected by the second reference position as a second reference signal; combining a plurality of echo signals transmitted by a second training position set according to time arrangement to obtain a second training signal set; the second reference signal and the second training signal constitute a second sample set.
Further, the specific step of acquiring the two-dimensional range-doppler image sets corresponding to the first sample set and the second sample set includes:
performing Fourier transformation on different echo signals in the first sample set and the second sample set to obtain a one-dimensional matrix of a distance dimension, setting the one-dimensional matrix obtained by the echo signals acquired for the first time as a first row of a two-dimensional matrix N, setting the one-dimensional matrix obtained by the echo signals acquired for the second time as a second row of the two-dimensional matrix N, and so on to obtain a one-dimensional matrix obtained by the echo signals acquired for the Nth time as an Nth row of the two-dimensional matrix N;
performing Fourier transform on the column vector of the two-dimensional matrix N to obtain another two-dimensional matrix M; the two-dimensional matrix M represents a two-dimensional distance-Doppler image corresponding to one frame of data acquired by the millimeter wave radar;
performing peak spectrum search on the two-dimensional distance-Doppler image to obtain a peak point of the two-dimensional distance-Doppler image and a corresponding numerical value Fij, i =1, 2, 3, ·, n of the two-dimensional distance-Doppler image; j =1, 2, 3, ·, n; i is the row coordinate of the peak point, j is the longitudinal coordinate of the peak point, and n is a positive integer;
taking i, j and Fij in a plurality of peak points in each two-dimensional distance-Doppler image as row vectors, forming a new matrix P by the row vectors, obtaining the maximum value of the third row in the matrix P, marking the row vector where the maximum value is located as a characteristic peak point, and so on to obtain the characteristic peak points in the two-dimensional distance-Doppler images; the characteristic peak point represents the movement information of the corresponding human hand in a period of time, and the movement information comprises the distance between the human hand and the millimeter wave radar and the radial speed of the human hand relative to the millimeter wave radar in the period of time;
sequentially connecting a plurality of characteristic peak points under a two-dimensional distance-Doppler coordinate system to obtain a trajectory diagram of hand movement; and the trajectory graph of the gesture is a moving trajectory of the human hand corresponding to the peak value of the characteristic point in a distance-Doppler coordinate system.
Further, the characteristic peak points on the first sample set and the second sample set trace graph are divided according to the preset dividing number, and a first dividing set corresponding to the first sample set and a second dividing set corresponding to the second sample set are obtained.
Further, the specific step of calculating the matching degree of the obtained difference frequency combination set and the difference frequency sample set includes:
acquiring data information of difference frequency signals in difference frequency combination set, wherein the data information comprises measuring speed, measuring distance and measuring angle related to the difference frequency signals, respectively acquiring numerical values of the measuring speed, the measuring distance and the measuring angle, marking the numerical values as CS, CJ and CD, carrying out normalization processing on various data with marked values, and carrying out normalization processing on the various data with the marked values through a formula
And calculating and obtaining the matching degree PP of the acquired difference frequency signal and the difference frequency signal of the sample in the difference frequency sample set, wherein YSP, YJP and JDP are the average measuring speed, the average measuring distance and the average measuring angle of the sample difference frequency signal respectively, and a1, a2 and a3 are different proportionality coefficients.
Further, the specific steps of comparing and screening the matching range formed by the matching degree and the difference frequency sample set comprise:
obtaining a matching range according to the difference frequency sample set, comparing and screening the matching degree and the matching range, if the matching degree belongs to the difference frequency sample range, judging that the difference frequency signals corresponding to the matching degree are in the identification range and generating identification instructions, and combining echo signals associated with a plurality of difference frequency signals according to the identification instructions in a time sequencing manner to obtain a selected signal set;
if the matching degree does not belong to the range of the difference frequency samples, judging that the difference frequency signals corresponding to the matching degree are not in the recognition range, generating a prompt instruction, prompting that the gesture is not standard according to the prompt instruction, and re-recognizing the gesture.
Further, acquiring the number of sets divided in the recognition sorting set and characteristic peak points in the sets, respectively matching the sets with the characteristic peak points divided in the first dividing set and the second dividing set, matching the characteristic peak points in the divided sets with the same proportion and setting the matching proportion, and analyzing the matching proportion;
if at least k matching proportions corresponding to the sets divided in the recognition sorting set are not smaller than a preset standard proportion, and k is a positive integer, judging that the gesture matching degree is high, generating a first recognition signal, and acquiring a gesture type corresponding to the recognition sorting set according to the first recognition signal;
if no k matching proportions corresponding to the sets divided in the recognition sorting set are not smaller than a preset standard proportion, judging that the gesture matching degree is medium and generating a second recognition signal, marking the gesture action as a correction action according to the second recognition signal, and acquiring data corresponding to the correction action to optimize and update the recognition database.
Further, the gesture action is marked as a correction action according to the second recognition signal, data corresponding to the correction action are obtained, the data comprise echo signals corresponding to the correction action and the position of the human hand, the echo signals and the position of the human hand are marked as optimized echo signals and optimized positions respectively, and the optimized echo signals and the optimized positions are operated through S1-S3 to achieve updating optimization of the recognition database.
The invention has the beneficial effects that:
according to the invention, a sample gesture recognition database is constructed in advance, so that data support can be provided for subsequent gesture recognition and gesture updating, echo signals are processed and analyzed, whether the echo signals of gesture recognition belong to the recognition range of the recognition database or not is judged, the echo signals which do not belong to the recognition range are not trained and recognized, the problem that the overall gesture recognition effect is influenced by processing of invalid data is avoided, and the overall efficiency of gesture recognition is effectively improved.
In the invention, the screened echo signals are matched and identified by the identification database and the identification result is output, the matching degree of gesture identification can be analyzed in the identification process, whether the matching degree of the gesture identification meets the requirement of optimization updating or not is judged, the identification database can be optimized and updated according to gesture actions with moderate matching degree, and the identification efficiency of the identification database is further improved.
Drawings
FIG. 1 is a flow chart of a millimeter wave radar micro-motion gesture recognition method according to the invention.
Detailed Description
Referring to fig. 1, the invention relates to a millimeter wave radar micro-motion gesture recognition method, which comprises the following specific steps:
s1: constructing a recognition database of sample gestures, wherein the recognition database comprises a first sample set and a second sample set of different states of the hand of the human body; the first sample set comprises various data of a mobile test of a human hand in a vertical state, and specifically comprises a first reference signal and a first training signal set; the second sample set comprises various data of the movement test of the human hand in a horizontal state, and specifically comprises a second reference signal and a second training signal set;
when the human hand is in a vertical state, setting the position of the human hand when the human hand is not moved as a first reference position, arranging and combining different positions of the human hand after moving according to time to obtain a first training position set, transmitting a test signal through a millimeter wave radar, receiving an echo signal reflected by the human hand through a signal receiver, and setting the echo signal reflected by the first reference position as a first reference signal; combining a plurality of echo signals transmitted by a first training position set according to time arrangement to obtain a first training signal set; the first reference signal and the first training signal set constitute a first sample set;
when the hands of the human body are in a horizontal state, setting the position of the hands of the human body when the hands of the human body are not moved as a second reference position, arranging and combining different positions of the hands of the human body after moving according to time to obtain a second training position set, transmitting a test signal through a millimeter wave radar, receiving an echo signal reflected by the hands of the human body through a signal receiver, and setting the echo signal reflected by the second reference position as a second reference signal; combining a plurality of echo signals transmitted by a second training position set according to time arrangement to obtain a second training signal set; the second reference signal and the second training signal constitute a second sample set;
in this embodiment, when the identification database of the sample gesture is constructed, the test signal transmitted by the millimeter wave radar is the same as the formally identified transmitted chirp signal, and both the test signal and the formally identified transmitted chirp signal can be realized by the same 77GHz LFMCW millimeter wave radar; by classifying and identifying the human hands in different states, an identification database can be established more efficiently and comprehensively so as to identify the gesture movement of the human hands more accurately;
it should be noted that the gesture movement includes, but is not limited to, the front and back, the left and right, and the up and down movement in the recognition distance range in the vertical state of the human hand, and the front and back, the left and right, and the up and down movement in the recognition distance range in the horizontal state of the human hand, the recognition distance range may be set in advance, the accuracy of the gesture recognition may be improved, in this embodiment, the maximum length of the recognition distance may be 1.2m, the minimum length of the recognition distance may be 0.3m, that is, the distance that can be recognized in the horizontal direction is 0.9m, the distance that can be recognized in the horizontal direction is 0.2m, the up and down height in the vertical direction may be 0.2m, that is, the distance that can be recognized in the vertical direction is 0.4 m.
S2: performing difference frequency processing and ADC (analog to digital converter) sampling on the test signal and the first sample set and the second sample set respectively to obtain a difference frequency sample set containing sample difference frequency signals;
in the embodiment, the difference frequency sample set is used for obtaining a difference frequency sample range, the collected difference frequency signal set can be matched and screened through the difference frequency sample range, whether the identified echo signal belongs to the echo signal range of sample training is judged, whether the identified gesture type is in the identification range is judged based on the echo signal, and when the gesture type is not in the identification range, prompting is performed without subsequent identification operation, and compared with the existing scheme of performing a series of processing, training and matching on the echo signal and prompting gesture identification failure, the method can effectively improve the identification efficiency and save the consumption of data resources;
s3: acquiring two-dimensional distance-Doppler image sets corresponding to a first sample set and a second sample set, and respectively processing and dividing to obtain a first division set corresponding to the first sample set and a second division set corresponding to the second sample set; the method comprises the following specific steps:
performing Fourier transformation on different echo signals in the first sample set and the second sample set to obtain a one-dimensional matrix of a distance dimension, setting the one-dimensional matrix obtained by the echo signals acquired for the first time as a first row of a two-dimensional matrix N, setting the one-dimensional matrix obtained by the echo signals acquired for the second time as a second row of the two-dimensional matrix N, and so on to obtain a one-dimensional matrix obtained by the echo signals acquired for the Nth time as an Nth row of the two-dimensional matrix N;
performing Fourier transform on the column vector of the two-dimensional matrix N to obtain another two-dimensional matrix M; the two-dimensional matrix M represents a two-dimensional distance-Doppler image corresponding to one frame of data acquired by the millimeter wave radar;
performing peak spectrum search on the two-dimensional distance-Doppler image to obtain a peak point of the two-dimensional distance-Doppler image and a corresponding numerical value Fij, i =1, 2, 3, ·, n of the two-dimensional distance-Doppler image; j =1, 2, 3, ·, n; i is the row coordinate of the peak point, j is the longitudinal coordinate of the peak point, and n is a positive integer;
taking i, j and Fij in a plurality of peak points in each two-dimensional distance-Doppler image as row vectors, forming a new matrix P by the row vectors, obtaining the maximum value of the third row in the matrix P, marking the row vector where the maximum value is located as a characteristic peak point, and so on to obtain the characteristic peak points in the two-dimensional distance-Doppler images; the characteristic peak point represents the movement information of the corresponding human hand in a period of time, and the movement information comprises the distance between the human hand and the millimeter wave radar and the radial speed of the human hand relative to the millimeter wave radar in the period of time;
sequentially connecting a plurality of characteristic peak points under a two-dimensional distance-Doppler coordinate system to obtain a trajectory diagram of hand movements; the trajectory graph of the gesture is a moving trajectory of the human hand corresponding to the peak value of the characteristic point in a distance-Doppler coordinate system;
specifically, a first column and a second column of each characteristic peak point are respectively used as an abscissa and an ordinate in sequence, adjacent characteristic peak points are marked and connected in sequence, and finally an image is obtained, wherein the image is a track graph of a gesture;
dividing characteristic peak points on the first sample set and the second sample set locus diagram according to a preset dividing number to obtain a first dividing set corresponding to the first sample set and a second dividing set corresponding to the second sample set;
s4: transmitting linear frequency modulation signals through a millimeter wave radar, receiving echo signals reflected by a hand of a human body through a signal receiver, and acquiring a difference frequency combination set through difference frequency by using a transmitting signal set formed by the linear frequency modulation signals and an echo signal set formed by the echo signals;
s5: calculating and obtaining the matching degree of the difference frequency combination set and the difference frequency sample set, comparing and screening the matching range formed by the matching degree and the difference frequency sample set, and setting the echo signal set meeting the standard as a selected signal set; the difference frequency sample set is a set which is obtained by training sample gestures and contains a plurality of sample difference frequency signals; the matching degree is used for analyzing and judging whether the difference frequency signal meets the requirement of gesture recognition; the method comprises the following specific steps:
acquiring data information of difference frequency signals in difference frequency combination set, wherein the data information comprises measuring speed, measuring distance and measuring angle related to the difference frequency signals, respectively acquiring numerical values of the measuring speed, the measuring distance and the measuring angle, marking the numerical values as CS, CJ and CD, carrying out normalization processing on various data with marked values, and carrying out normalization processing on the various data with the marked values through a formula
Calculating and obtaining the matching degree PP of the acquired difference frequency signals and the difference frequency signals of the samples in the difference frequency sample set, wherein YSP, YJP and JDP are the average measuring speed, the average measuring distance and the average measuring angle of the sample difference frequency signals respectively, and a1, a2 and a3 are different proportionality coefficients;
obtaining a matching range according to the difference frequency sample set, comparing and screening the matching degree and the matching range, if the matching degree belongs to the difference frequency sample range, judging that the difference frequency signals corresponding to the matching degree are in the identification range and generating identification instructions, and combining echo signals associated with a plurality of difference frequency signals according to the identification instructions in a time sequencing manner to obtain a selected signal set; the matching range is set according to the minimum change speed, the minimum change distance and the minimum change angle in the difference frequency sample set, the maximum change speed, the maximum change distance and the maximum change angle respectively, the average measurement speed, the average measurement distance and the average measurement angle through the formula, and the minimum value and the maximum value of the matching range are obtained;
if the matching degree does not belong to the range of the difference frequency samples, judging that the difference frequency signals corresponding to the matching degree are not in the recognition range, generating a prompt instruction, prompting that the gesture is not standard according to the prompt instruction, and re-recognizing the gesture;
in the embodiment, the difference frequency signal is associated with the linear frequency modulation signal and the received echo signal, a cylindrical coordinate system is established for the millimeter wave radar, the position of the hand of the human body is positioned through the horizontal distance, the azimuth angle and the height, and the measuring speed, the measuring distance and the measuring angle are obtained according to the position change of the hand of the human body; the gesture recognition is not normative, including but not limited to the recognized distance, the recognized speed, and the type of motion exceeding the recognition range of the recognition database, and the gesture recognition needs to be performed again.
S6: filtering and monitoring echo signals in the selected signal set to obtain an echo monitoring set, acquiring a two-dimensional distance-Doppler image set corresponding to the echo monitoring set and setting the two-dimensional distance-Doppler image set as an identification image set, acquiring characteristic peak point coordinates of a trace map corresponding to the identification image set and sequencing the characteristic peak point coordinates according to time to obtain an identification sequencing set;
dividing the identification sorting set according to a preset division number to obtain an identification division set;
in this embodiment, an acquisition scheme of a two-dimensional range-doppler image set corresponding to the echo monitoring set is the same as an acquisition scheme of a two-dimensional range-doppler image set corresponding to the first sample set and the second sample set; the identification sorting set is divided, so that the modular analysis of the identification gestures can be realized, and the accuracy of gesture identification analysis can be improved;
it should be noted that, when the filtering process is performed under the butterworth low-pass filter, the number of the two-dimensional range-doppler images in the two-dimensional range-doppler image set may be 100, and then the number of the corresponding characteristic peak points is also 100, and the preset number of the partitions may be 10, and then the 100 characteristic peak points are equally divided into 10 parts; the purpose of partitioning the recognition ordered set is to match and optimize the gesture for subsequent more precision.
S7: matching and recognizing the recognition sorting set and a recognition database of the pre-constructed sample gestures to obtain a recognition result of the human body gestures, and optimizing and updating the pre-constructed recognition database according to the recognition result; the method comprises the following specific steps:
acquiring the number of sets divided in the recognition sorting set and the characteristic peak points in the sets, respectively matching the sets with the characteristic peak points divided in the first dividing set and the second dividing set, matching the characteristic peak points in the divided sets with the same proportion and setting the matching proportion as a matching proportion, and analyzing the matching proportion;
if at least k matching proportions corresponding to the sets divided in the recognition sorting set are not smaller than a preset standard proportion, and k is a positive integer, judging that the gesture matching degree is high, generating a first recognition signal, and acquiring a gesture type corresponding to the recognition sorting set according to the first recognition signal;
if no k matching proportions corresponding to the sets divided in the recognition sorting set are not smaller than a preset standard proportion, judging that the gesture matching degree is medium and generating a second recognition signal, marking the gesture action as a correction action according to the second recognition signal, and acquiring data corresponding to the correction action to perform optimization updating on a recognition database; the method comprises the following specific steps:
and acquiring data corresponding to the correction action, wherein the data comprises an echo signal corresponding to the correction action and the position of the human hand, marking the echo signal and the position of the human hand as an optimized echo signal and an optimized position respectively, and realizing updating optimization of the identification database by operating the optimized echo signal and the optimized position through S1-S3.
The gestures associated with the selected signal set after the operation of S5 can be recognized normally, but the recognition accuracy needs to be further determined so as to optimize the recognition database better; in this embodiment, the preset standard proportion may be 90%, and the value of k may be 8, that is, in 10 parts of 100 feature peak points equally divided, when at least 90% of feature peak points in each part are the same and at least 8 parts are the same, it is determined that the gesture matching degree is high, and if gesture recognition that does not reach the standard is not achieved, it is determined that the gesture matching degree is medium, and it is necessary to update and optimize the recognition database.
The formula in the invention is a formula which is obtained by removing dimensions, taking the numerical value of the dimension to calculate, and acquiring a large amount of data to perform software simulation to obtain the closest real condition, wherein the preset proportionality coefficient in the formula is set by a person skilled in the art according to the actual condition or acquired through simulation of a large amount of data.
The above description is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments, and all technical solutions belonging to the idea of the present invention belong to the protection scope of the present invention. It should be noted that modifications and adaptations to those skilled in the art without departing from the principles of the present invention should also be considered as within the scope of the present invention.
Claims (7)
1. A millimeter wave radar micro-motion gesture recognition method is characterized by comprising the following steps:
transmitting a linear frequency modulation signal through a millimeter wave radar, receiving an echo signal reflected by a hand of a human body through a signal receiver, and acquiring a difference frequency combination set according to a transmitting signal set formed by the linear frequency modulation signal and an echo signal set formed by the echo signal;
calculating and obtaining the matching degree of the difference frequency combination set and the difference frequency sample set, comparing and screening the matching range formed by the matching degree and the difference frequency sample set, and setting the echo signal set meeting the standard as a selected signal set; the difference frequency sample set is a set which is obtained by training sample gestures and contains a plurality of sample difference frequency signals; the matching degree is used for analyzing and judging whether the difference frequency signal meets the requirement of gesture recognition;
the specific steps of calculating and obtaining the matching degree of the difference frequency combination set and the difference frequency sample set comprise:
acquiring data information of difference frequency signals in difference frequency combination set, wherein the data information comprises measuring speed, measuring distance and measuring angle related to the difference frequency signals, respectively acquiring numerical values of the measuring speed, the measuring distance and the measuring angle, marking the numerical values as CS, CJ and CD, and marking each item of the numerical values as CS, CJ and CD
Normalizing the data by formula
Calculating and obtaining the matching degree PP of the acquired difference frequency signals and the difference frequency signals of the samples in the difference frequency sample set, wherein YSP, YJP and JDP are the average measuring speed, the average measuring distance and the average measuring angle of the sample difference frequency signals respectively, and a1, a2 and a3 are different proportionality coefficients;
the specific steps of comparing and screening the matching range formed by the matching degree and the difference frequency sample set comprise:
obtaining a matching range according to the difference frequency sample set, comparing and screening the matching degree and the matching range, if the matching degree belongs to the matching range, judging that the difference frequency signals corresponding to the matching degree are in the identification range and generating identification instructions, and sequencing and combining echo signals associated with a plurality of difference frequency signals according to the identification instructions to obtain a selected signal set;
filtering and monitoring echo signals in the selected signal set to obtain an echo monitoring set, processing and dividing the echo monitoring set to obtain a two-dimensional distance-Doppler image set corresponding to the echo monitoring set and setting the two-dimensional distance-Doppler image set as an identification image set, and obtaining characteristic peak point coordinates of a trace map corresponding to the identification image set and sequencing the characteristic peak point coordinates according to time to obtain an identification sequencing set; the trajectory graph of the gesture is a moving trajectory of the human hand corresponding to the peak value of the characteristic point in a distance-Doppler coordinate system; dividing the identification sorting set according to a preset division number to obtain an identification division set;
and matching and recognizing the recognition sequencing set and a recognition database of the pre-constructed sample gestures to obtain a recognition result of the human body gestures, and optimizing and updating the pre-constructed recognition database according to the recognition result.
2. The millimeter wave radar micro-motion gesture recognition method according to claim 1, wherein a recognition database of sample gestures is constructed, the recognition database comprising a first sample set and a second sample set of different states of a human hand; the first sample set comprises a first reference signal and a first training signal set; the second set of samples includes a second reference signal and a second set of training signals;
performing difference frequency processing and ADC (analog to digital converter) sampling on the test signal and the first sample set and the second sample set respectively to obtain a difference frequency sample set containing sample difference frequency signals;
and acquiring two-dimensional distance-Doppler image sets corresponding to the first sample set and the second sample set, and processing and dividing the two-dimensional distance-Doppler image sets respectively to obtain a first division set corresponding to the first sample set and a second division set corresponding to the second sample set.
3. The millimeter wave radar micro-gesture recognition method according to claim 2, wherein when the human hand is in a vertical state and a horizontal state, the positions where the human hand does not move in the vertical state and the horizontal state are respectively set as a first reference position and a second reference position, different positions of the human hand after moving in different states are combined according to time arrangement to obtain a first training position set and a second training position set, a test signal is transmitted by a millimeter wave radar, an echo signal reflected by the human hand is received by a signal receiver, and echo signals reflected by the first reference position and the second reference position are respectively set as a first reference signal and a second reference signal.
4. The millimeter wave radar micro-gesture recognition method according to claim 3, wherein a plurality of echo signals transmitted by the first training position set and the second training position set are respectively combined according to time arrangement to obtain a first training signal set and a second training signal set;
the first reference signal and the first set of training signals form a first set of samples and the second reference signal and the second set of training signals form a second set of samples.
5. The millimeter wave radar micro gesture recognition method according to claim 4, wherein a plurality of two-dimensional range-Doppler images are obtained according to a first reference signal and a first training signal set in a first sample set and a second reference signal and a second training signal set in a second sample set;
respectively searching the spectral peak of each two-dimensional distance-Doppler image and solving the characteristic peak point of the two-dimensional distance-Doppler image; sequentially connecting a plurality of characteristic peak points under a two-dimensional distance-Doppler coordinate system to obtain a trajectory diagram of hand movements; and dividing the characteristic peak points on the first sample set and the second sample set locus diagram according to the preset dividing number to obtain a first dividing set corresponding to the first sample set and a second dividing set corresponding to the second sample set.
6. The millimeter wave radar micro-motion gesture recognition method according to claim 5, wherein the specific steps of obtaining the recognition result of the human body gesture comprise:
acquiring the number of sets divided in the identification sorting set and the characteristic peak points in the sets, and respectively matching the characteristic peak points divided in the first dividing set and the second dividing set;
if at least k matching proportions corresponding to the sets divided in the recognition sorting set are not smaller than a preset standard proportion, and k is a positive integer, generating a first recognition signal and acquiring a gesture type corresponding to the recognition sorting set;
if no k matching proportions corresponding to the sets divided in the recognition sorting set are not smaller than a preset standard proportion, generating a second recognition signal, and optimizing the recognition database according to the second recognition signal; and the matching proportion is the proportion of matching the characteristic peak points in the divided set.
7. The millimeter wave radar micro-motion gesture recognition method according to claim 6, wherein the gesture motion is marked as a modification motion according to the second recognition signal, data corresponding to the modification motion is obtained, the data comprises an echo signal corresponding to the modification motion and a position of a human hand, the echo signal and the position of the human hand are respectively marked as an optimized echo signal and an optimized position, and the optimized echo signal and the optimized position are supplemented into the recognition database for updating and optimizing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210164177.XA CN114236492B (en) | 2022-02-23 | 2022-02-23 | Millimeter wave radar micro gesture recognition method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210164177.XA CN114236492B (en) | 2022-02-23 | 2022-02-23 | Millimeter wave radar micro gesture recognition method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114236492A CN114236492A (en) | 2022-03-25 |
| CN114236492B true CN114236492B (en) | 2022-06-07 |
Family
ID=80747769
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210164177.XA Active CN114236492B (en) | 2022-02-23 | 2022-02-23 | Millimeter wave radar micro gesture recognition method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114236492B (en) |
Family Cites Families (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3133474B1 (en) * | 2015-08-19 | 2019-03-27 | Nxp B.V. | Gesture detector using ultrasound |
| CN110309690B (en) * | 2018-03-27 | 2022-09-27 | 南京理工大学 | Gesture recognition detection method based on time frequency spectrum and range-Doppler spectrum |
| CN111476058B (en) * | 2019-01-23 | 2024-05-14 | 北京邮电大学 | Gesture recognition method based on millimeter wave radar |
| CN112363156A (en) * | 2020-11-12 | 2021-02-12 | 苏州矽典微智能科技有限公司 | Air gesture recognition method and device and intelligent equipment |
| CN112416128B (en) * | 2020-11-23 | 2022-07-01 | 森思泰克河北科技有限公司 | Gesture recognition method and terminal equipment |
| CN113553923A (en) * | 2021-07-05 | 2021-10-26 | 镇江同润智能科技有限公司 | Gesture recognition method based on millimeter wave radar |
-
2022
- 2022-02-23 CN CN202210164177.XA patent/CN114236492B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN114236492A (en) | 2022-03-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Deep learning and recognition of radar jamming based on CNN | |
| CN107729916B (en) | ISODATA-based interference source classification and identification algorithm | |
| Liu | Multi-feature fusion for specific emitter identification via deep ensemble learning | |
| CN111461037B (en) | End-to-end gesture recognition method based on FMCW radar | |
| CN114449452B (en) | Wi-Fi indoor positioning method based on CNN-RNN | |
| CN112734809B (en) | On-line multi-pedestrian tracking method and device based on Deep-Sort tracking framework | |
| CN111813222B (en) | Terahertz radar-based fine dynamic gesture recognition method | |
| CN113050797A (en) | Method for realizing gesture recognition through millimeter wave radar | |
| CN116071600B (en) | Crop remote sensing identification method and device based on multi-classification probability | |
| CN113762130B (en) | Millimeter wave radar gesture detection and recognition method | |
| CN111476158A (en) | Multi-channel physiological signal somatosensory gesture recognition method based on PSO-PCA-SVM | |
| CN114236492B (en) | Millimeter wave radar micro gesture recognition method | |
| CN103955711A (en) | Mode recognition method in imaging spectrum object recognition analysis | |
| CN113923589B (en) | Non-contact type multi-target positioning method | |
| CN115909086A (en) | SAR target detection and identification method based on multistage enhanced network | |
| CN111062322A (en) | Phased array radar behavior recognition method based on Support Vector Machine (SVM) | |
| CN111860616B (en) | General acquisition method for weak contrast collimation image target center of comprehensive diagnosis system | |
| CN113030849B (en) | Near field source positioning method based on self-encoder and parallel network | |
| CN116400724A (en) | Intelligent inspection method for unmanned aerial vehicle of power transmission line | |
| CN111860613B (en) | Multi-target tracking and state predicting method based on multi-source heterogeneous signals | |
| CN111239682A (en) | Electromagnetic emission source positioning system and method | |
| CN110880012A (en) | Correlation method for frequency information of agile radar radiation source between multiple scout platforms | |
| CN110110115B (en) | Method for screening and identifying airborne target image through man-machine interaction | |
| CN113259847B (en) | Indoor positioning method based on feature optimization and LightGBM | |
| CN108111240A (en) | The data processing method and device of a kind of radio frequency interference |
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 | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240617 Address after: 212000 No. 103, Zhongshan West Road, Runzhou District, Zhenjiang City, Jiangsu Province Patentee after: Zhenjiang Tongrun Intelligent Technology Co.,Ltd. Country or region after: China Address before: 1901, 6th floor, Guoyao building, No. 6 Hanzhong Road, Gulou District, Nanjing, Jiangsu 210003 Patentee before: Nanjing Yichun Technology Co.,Ltd. Country or region before: China Patentee before: Zhenjiang Tongrun Intelligent Technology Co.,Ltd. |