CN105989710B - A kind of device for monitoring vehicle and method based on audio - Google Patents
A kind of device for monitoring vehicle and method based on audio Download PDFInfo
- Publication number
- CN105989710B CN105989710B CN201510073084.6A CN201510073084A CN105989710B CN 105989710 B CN105989710 B CN 105989710B CN 201510073084 A CN201510073084 A CN 201510073084A CN 105989710 B CN105989710 B CN 105989710B
- Authority
- CN
- China
- Prior art keywords
- vehicle
- lane
- detection zones
- coarseness
- module
- 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
Abstract
A kind of device for monitoring vehicle based on audio, comprising: microphone array module: for acquiring and handling the noise signal of vehicle sending, the correlation matrix C of horizontal and vertical subarray is obtained;Lane position and width computing module: for calculating position and the width in each lane;Coarseness detection zone energy spectrum computing module: the energy spectrum in two coarseness detection zones is calculated for constructing two coarseness detection zones on each lane, and using correlation matrix C;Automatic growth control module: for calculating prospect threshold alpha and background threshold β, the energy spectrum in two coarseness detection zones is normalized, and judges whether vehicle passes in and out two coarseness detection zones;Vehicle count module: the vehicle number passed through for counting each lane;Lane occupancy ratio computing module: for calculating the occupation rate in each lane;Speed estimation module: for estimating the speed of vehicle;Vehicle classification module: size type used for vehicles is classified.
Description
Technical field
The invention belongs to field of intelligent transportation technology, and in particular to a kind of device for monitoring vehicle and method based on audio.
Background technique
In modern road traffic transportation system process of construction, it will usually dispose a large amount of vehicle monitoring on roadside
Equipment enables control centre to be monitored in real time to the traffic condition on each road, and is planned according to different road conditions
Traffic control measure.As " eyes " and " ear " of control centre, various vehicle monitoring sensors are in intelligent transportation system
It plays an important role.It is made more rationally so more advanced, reliable, the accurate monitoring device of research and development will be helpful to people
Traffic control decision, and then reduce congestion in road, reduce transportation cost, and improve the safety of highway communication.
For different road environments and monitoring demand, people have developed different types of vehicle monitoring equipment.Wherein,
Magnetic induction loop is a kind of monitoring device of classics, has been continued to use so far more than 50 years.But this kind of equipment need to be embedded in road surface
Lower section, so high installation and maintenance cost can be generated.It is different from magnetic induction loop, the installation and dimension of modern vehicle monitoring device
Shield mostly all will not road pavement damage.According to its principle difference, can by modern vehicle monitoring device it is rough to be divided into two big
Class: active and passive type.Active monitoring device is based primarily upon the technologies such as laser, infrared, radar, ultrasound, and this kind of equipment is logical
It crosses and a kind of specific signal is emitted to monitoring area and receives its echo, then vehicle is determined to realize according to the characteristics of echo
Position and monitoring;Passive type equipment is based primarily upon the technologies such as the infrared, audio processing of video, passive type, such equipment directly passes through vehicle
The signal (such as visible light) of reflection or the signal (such as infrared, noise) issued realize the purpose of monitoring.
It is compared with other methods, the monitoring method based on audio has many advantages: firstly, compared to other kinds of prison
It controls for sensor, the cost of audio sensor (microphone) wants much lower, and Voice Surveillance belongs to passive type monitoring, is not required to
Sender unit is wanted, this advantageously reduces the cost of entire monitoring device;Second, the feature based on audio is not readily susceptible to day
The influence of the environmental factors such as gas, illumination, this is conducive to system and realizes round-the-clock stable vehicle monitoring.It can however, developing one kind
There is also many difficult points for the audio device for monitoring vehicle leaned on: firstly, since the wavelength of automobile noise is located at centimetre this quantity
Grade, this is compared to (such as the grade of radar, it is seen that the micron of light of signal wavelength used in other kinds of monitoring device
Grade) it is much greater, so, the resolution ratio of Voice Surveillance device is lower than the method based on other signals in theory;The
Two, due to also containing the sound of various other types in external environment other than automobile noise, and on real road often
There are more vehicles to travel at the same time, that is, needs to handle more sound sources, the situation with noise source, and so complexity external environment is passed through vapour
Vehicle noise realizes that vehicle monitoring increases difficulty.So needing prominent use in the development process of audio device for monitoring vehicle
The advantages of audio signal is as medium, while overcoming its deficiency, to achieve the purpose that maximize favourable factors and minimize unfavourable ones.
Research and applicating history of the audio vehicle monitoring in external existing more than ten years, United States Patent (USP) US 5,798,983
(Acoustic Sensor System for Vehicle Detection and Multi-Lane Highway
Monitoring, John Patrick Kuhn) use the microphone array array structure based on square matrix, but the array element of the patent
Quantity is on the high side, and the resolution ratio of its lane location algorithm is not high.United States Patent (USP) US 6,195,608B1 (Acoustic
Highway Monitor, Edward Fredrick Berliner) array structure based on cross battle array is used, although the patent
Array number is reduced, but cannot estimate speed.
The country starts late to the research and application of audio vehicle monitoring, the Chinese patent of Wu Xihong et al.
CN102682765B, expressway audio vehicle detection device and its method.The basic principle of this method determines that needs will be right
The vehicle detection apparatus answered is installed on portal frame, and installation cost is higher;In addition, this method is not suitable for the more ring of vehicle
Border.
Summary of the invention
It is an object of the invention to overcome drawbacks described above existing for current audio vehicle monitoring system, proposes one kind and be based on
The device for monitoring vehicle of audio, and propose a kind of vehicle monitoring method based on audio based on the device, this method can be
Two coarseness detection zones are constructed on lane, realize the statistics of the various index parameters to reflection traffic condition.
To achieve the goals above, the present invention provides a kind of device for monitoring vehicle based on audio, comprising:
Microphone array module: for acquiring and handling the noise signal of vehicle sending;Each array element in microphone array
The vehicle noise time-domain signal of acquisition forms the time domain data of horizontal and vertical subarray after analog-to-digital conversion;To time domain number
According to multichannel Short Time Fourier Transform is carried out, its frequency domain data is obtained;Sub-band and time interval is selected to calculate laterally and vertical
To the correlation matrix C of subarray;
The microphone array takes cross array structure;Horizontal and vertical subarray all uses an equidistant word array
Structure, and the array number of two subarrays is different with array element spacing;The low-limit frequency f for the signal that the array can be handledminIt is greater than
Equal to 7kHz;And laterally and longitudinally the array element spacing of subarray should meet dh<λmin/ 2, dv<λmin, wherein λminMost for signal
Small wavelength;Thereby determine that microphone array can handle the maximum frequency f of signalmax, lateral subarray array number MhAnd array element
Spacing dh, longitudinal subarray array number MvWith array element spacing dv;
Lane position and width computing module;For calculating position and the width in each lane;
Coarseness detection zone energy spectrum computing module: for constructing two coarseness detection zones on each lane;
And the energy spectrum in two coarseness detection zones is calculated using correlation matrix C;
Automatic growth control module: for calculating prospect threshold alpha and background threshold β, in two coarseness detection zones
Energy spectrum be normalized, and judge whether vehicle passes in and out two coarseness detection zones;
Vehicle count module: the vehicle number passed through for counting each lane;
Lane occupancy ratio computing module: for calculating the occupation rate in each lane;
Speed estimation module: for estimating the speed of vehicle;
Vehicle classification module: size type used for vehicles is classified.
Based on the above-mentioned device for monitoring vehicle based on audio, the present invention also provides a kind of vehicle monitoring sides based on audio
Method, which comprises
Step 1) starts microphone array module, the vehicle noise time-domain signal warp of each array element acquisition in microphone array
After crossing analog-to-digital conversion, the time domain data of horizontal and vertical subarray is formed;Carrying out multichannel to time domain data, Fourier becomes in short-term
It changes, obtains its frequency domain data;Selection sub-band and time interval calculate the correlation matrix C of horizontal and vertical subarray;
Step 2) starts lane position and width computing module, calculates position and the width in each lane;
Position and the width for obtaining the lane of the monitored highway of the device for monitoring vehicle, utilize the peace of device for monitoring vehicle
Dress height, at elevation angle theta=0 °, calculates the corresponding azimuthal range in each lane:
Step 3) starts coarseness detection zone energy spectrum computing module, and two coarseness detections are constructed on each lane
Region, and the energy spectrum in two coarseness detection zones is calculated using correlation matrix C;
Step 4) starts automatic growth control module, calculates prospect threshold alpha and background threshold β;And two coarsenesses are examined
The energy spectrum surveyed on region is normalized;Judge whether vehicle passes in and out two coarseness detection zones;
Step 5) starts vehicle count module, calculates vehicle flowrate;
When having sequence of cars after two coarseness detection zones in a lane, add for the vehicle flowrate counting in the lane
1;
Step 6) starts lane occupancy ratio computing module, calculates lane occupancy ratio;
Step 7) starts speed estimation module, estimates speed;
Step 8) starts vehicle classification module, and the classification of large and small vehicle is carried out to vehicle.
In above-mentioned technical proposal, the step 1) further comprises:
The time-domain signal of each array element acquisition vehicle noise of step 101) microphone array, obtains after analog-to-digital conversion
The time domain data x of lateral subarray and longitudinal subarrayh(t) and xv(t);
Step 102) is to time domain data xh(t) and xv(t) multichannel Short Time Fourier Transform is carried out, frequency domain data x is obtainedh
(f, τ) and xv(f, τ), wherein f is frequency range number, and τ is the serial number of data frame;
Step 103) selects sub-band [f1,f2] and the corresponding correlation of the horizontal and vertical subarray of time interval Δ τ calculating
Matrix C:
Wherein, fmin<f1<f2<fmax, subscript H expression conjugate transposition.
In above-mentioned technical proposal, the step 3) further comprises:
Step 301) utilizes the width in laneConstruct ideal array response
Step 302) solves the Beam-former w of single coarseness detection zone using least square methodh;
A=[ah(-90),ah(-89),…,ah(0),…,ah(89),ah(90)]T
Beam-former whMeet:
It is solved using least square method:
Step 303) is that each lane constructs two coarseness detection zones, calculates the energy in two coarseness detection zones
Amount spectrum;
Beam-former whWhat is controlled is the width of coarseness detection zone, and the elevation angle of coarseness detection zone is then by indulging
To the Beam-former w of subarrayv(θ) is controlled;Two coarseness detection zones are constructed in every lane, the two regions
The elevation angle is respectively θ=- δ and θ=δ, wherein δ=5 °;The energy spectrum of two coarseness detection zones is respectively as follows:
In above-mentioned technical proposal, the step 4) further comprises:
Step 401) calculates prospect threshold alpha and background threshold β;
The energy spectrum of two coarseness detection zones is normalized in step 402);
Step 403) judges whether vehicle passes in and out two coarseness detection zones;
WhenWhen, judgement has vehicle to enter first coarseness detection zone;WhenWhen, judgement have vehicle into
Enter second coarseness detection zone;WhenAndWhen, judge that no vehicle enters coarseness detection zone;Wherein
γ is first threshold.
In above-mentioned technical proposal, the step 401) further comprises:
Step a) brings into operation from system, every 2 seconds, takes q in the time window1And q2Local maximum be
The method that step b) is estimated using the gloomy window of pa, by eachFit the vehicle noise energy that systematic observation obtains
Probability density curve;
Step c) determines the corresponding energy spectrum size of second wave crest in the probability density curve, is prospect threshold alpha;
It determines energy spectrum size corresponding to the trough between first and second wave crests, is background threshold β.
In above-mentioned technical proposal, the specific implementation process of the estimation speed of the step 7) are as follows:
Enter the time difference of two coarseness detection zones by vehicle to estimate speed v:
Wherein t1And t2Respectively vehicle enter a lane two coarseness detection zones time, h be the vehicle
Distance of the monitoring device to the lane;H is determined by the height of lane position, lane width and monitoring device installation.
In above-mentioned technical proposal, the specific implementation process of the classification of large and small vehicle is carried out in the step 8) to vehicle are as follows:
The calculation method of Vehicle length are as follows:
L=(tout-tin)v
Wherein, tinAnd toutRespectively vehicle enters and is driven out to time of a coarseness detection zone;
As l >=l0When, judge to pass through vehicle as large car;It otherwise is compact car, wherein l0For second threshold.
Compared with the prior art, the advantages of the present invention are as follows:
1, device for monitoring vehicle of the invention need to only be installed in infrastructure by the road, be easily installed and safeguard;
2, the feature that vehicle monitoring method of the invention has calculation amount small, low to hardware requirement.
Detailed description of the invention
Fig. 1 is coarseness detection zone schematic diagram of the invention;
Fig. 2 is scheme of installation of the device of the invention in road environment;
Fig. 3 is the structure composition schematic diagram of the device for monitoring vehicle of the invention based on audio;
Fig. 4 is the formation schematic diagram of microphone array of the invention.
Specific embodiment
Basic principle of the invention is multiple out in different placement configurations using Wave beam forming (Beamforming) technology
Vehicle detection region.Due to the spa-tial filter properties of Wave beam forming, the vehicle noise except detection zone will be by wave
Beam formation algorithm inhibits, so, can the signal as obtained by cumulative energy to determine whether there is vehicle to be located at some detection zone
In domain.For vehicle on highway monitoring, shape, size, position difference can be constructed according to different monitoring purposes
Multiple vehicle detection regions.As shown in Figure 1, device for monitoring vehicle is installed on roadside, it is that every lane constructs two in figure
Coarseness detection zone.Coarseness detection zone is used for the vehicle detection in each lane, and the width of this kind of detection zone is across whole
A lane, and be responsible on each lane by two coarseness detection zones.The parameter of common reflection traffic condition, such as wagon flow
Amount, lane occupancy ratio can be gone out by counting the number of vehicles and time reckoning in region after testing, and speed and vehicle are (big
Vehicle, trolley) it can then calculate the time difference by vehicle by two coarseness detection zones in same lane.Due to lane
Position may change with the variation of weather, temperature and traffic condition, so the position in coarseness region and width
Also it can change with lane position.
As shown in Figure 1, coordinate system used in the present invention is spherical coordinate system, the origin of coordinate system is in microphone array
The heart, abscissa are azimuthOrdinate is elevation angle theta, the orientation of vehicle byIt is indicated with θ.Due to the normal side of microphone array
To the center for being directed toward road surface, the azimuth of vehicleRange from -90 ° to 90 °, 0 ° of azimuth is located at the center on road surface, -90 °
Close to the rightmost of highway, 90 ° of Far Lefts close to highway.
The present invention is described further in the following with reference to the drawings and specific embodiments.
As shown in Fig. 2, device for monitoring vehicle of the invention is installed in existing wayside infrastructure, the normal side of equipment
The road-center position monitored is needed to direction.Vehicle can issue noise when driving on road and (make an uproar comprising engine noise, tire
The ingredients such as sound, exhaust noise, aerodynamic noise), vehicle noise signal is captured and is handled by Voice Surveillance device, various for calculating
Traffic statistics index, and then for monitoring road vehicle.
As shown in figure 3, a kind of device for monitoring vehicle based on audio includes:
Microphone array module: for acquiring and handling the noise signal of vehicle sending;Each array element in microphone array
The vehicle noise time-domain signal of acquisition forms the time domain data of horizontal and vertical subarray after modulus (A/D) conversion;Clock synchronization
Numeric field data carries out multichannel Short Time Fourier Transform, obtains its frequency domain data;Sub-band and time interval is selected to calculate laterally
With the correlation matrix of longitudinal subarray.
As shown in figure 4, the microphone array takes cross array structure;Horizontal and vertical subarray all uses equidistantly
Array structure, but the array number of two subarrays and array element spacing are different.
For multilane monitoring, it is desirable that equipment can distinguish the vehicle in adjacent lane driving alongside.Due to China
Highway lane width is 3.75 meters, and equipment mounting height is generally 10 meters, by these road environment parameters and according to wave
The basic theories that beam forms technology can estimate the array number and array element spacing of two subarrays.
For reasonable array aperture size, the low-limit frequency f of signalminIt cannot be below 7kHz;According to spatial sampling
Law, whenWhen, as θ ∈ [- δ, δ], wherein when δ≤5 °, in order to avoid occur space ambiguousness, laterally with
The array element spacing of longitudinal subarray should meet dh<λmin/ 2, dv<λmin, wherein λminFor the minimum wavelength of signal;In summary it limits
Condition, which can extrapolate microphone array, can handle the highest frequency f of signalmax, lateral subarray array number MhBetween array element
Away from dh, longitudinal subarray array number MvWith array element spacing dv。
In the present embodiment, horizontal aperture is having a size of 20cm, and longitudinal aperture is having a size of 32cm;δ=5 °;Take fmax=8kHz,
The array number M of lateral subarrayh=11, array element spacing dh=2cm, the array number M of longitudinal subarrayv=9, array element spacing dv=
4cm。
Microphone array does not only have a kind of above-mentioned embodiment, can be according to specific road environment and application demand pair
Array number and array element spacing are further optimized.
Lane position and width computing module: for calculating position and the width in each lane;
Coarseness detection zone energy spectrum computing module: for constructing two coarseness detection zones on each lane;
And the energy spectrum in two coarseness detection zones is calculated using correlation matrix C;
Automatic growth control module: for calculating prospect threshold alpha and background threshold β, in two coarseness detection zones
Energy spectrum be normalized, and judge whether vehicle passes in and out two coarseness detection zones;
Vehicle count module: the vehicle number passed through for counting each lane;
Lane occupancy ratio computing module: for calculating the occupation rate in each lane;
Speed estimation module: for estimating the speed of vehicle;
Vehicle classification module: size type used for vehicles is classified.
Based on the above-mentioned device for monitoring vehicle based on audio, the present invention provides a kind of vehicle monitoring sides based on audio
Method, which comprises
Step 1) starts microphone array module, the vehicle noise time-domain signal warp of each array element acquisition in microphone array
After crossing analog-to-digital conversion, the time domain data of horizontal and vertical subarray is formed;Carrying out multichannel to time domain data, Fourier becomes in short-term
It changes, obtains its frequency domain data;Selection sub-band and time interval calculate the correlation matrix C of horizontal and vertical subarray;Specifically
Include the following steps:
The time-domain signal of each array element acquisition vehicle noise of step 101) microphone array, obtains after A/D is converted
The time domain data x of lateral subarray and longitudinal subarrayh(t) and xv(t);
In the present embodiment, the sample rate of signal is 20kHz.
Step 102) is to time domain data xh(t) and xv(t) multichannel Short Time Fourier Transform is carried out, frequency domain data x is obtainedh
(f, τ) and xv(f, τ), wherein f is frequency range number, and τ is the serial number of data frame;
Step 103) selects sub-band [f1,f2] and the corresponding correlation of the horizontal and vertical subarray of time interval Δ τ calculating
Matrix C:
Wherein, fmin<f1<f2<fmax, subscript H expression conjugate transposition.
Correlation matrix C is the data of microphone array module output;It changes with the variation of time, with each battle array
The collected vehicle noise signal of member is constantly updated, and Matrix C can also be constantly updated, to ensure that the real-time of data.
Step 2) starts lane position and width computing module, calculates position and the width in each lane;
Position and the width for obtaining the lane of the monitored highway of the device for monitoring vehicle, utilize the peace of device for monitoring vehicle
Dress height, at elevation angle theta=0 °, calculates the corresponding azimuthal range in each lane:
Step 3) starts coarseness detection zone energy spectrum computational submodule, and two coarseness inspections are constructed on each lane
Region is surveyed, and calculates the energy spectrum in two coarseness detection zones using correlation matrix C;Specific steps are as follows:
Step 301) utilizes the width in laneConstruct ideal array response
Step 302) solves the Beam-former w of single coarseness detection zone using least square methodh;
A=[ah(-90),ah(-89),…,ah(0),…,ah(89),ah(90)]T
Beam-former whMeet:
It is solved using least square method:
Step 303) is that each lane constructs two coarseness detection zones, calculates the energy in two coarseness detection zones
Amount spectrum;
Beam-former whWhat is controlled is the width of coarseness detection zone, and the elevation angle of coarseness detection zone is then by indulging
To the Beam-former w of subarrayv(θ) is controlled;Two coarseness detection zones are constructed in every lane, the two regions
The elevation angle is respectively θ=- δ and θ=δ, and the energy spectrum of two coarseness detection zones is respectively as follows:
In the present embodiment, δ=5 °.
Step 4) starts automatic growth control module, calculates prospect threshold alpha and background threshold β;And two coarsenesses are examined
The energy spectrum surveyed on region is normalized;Judge whether vehicle passes in and out two coarseness detection zones;
Step 401) calculates prospect threshold alpha and background threshold β;Specifically comprise the following steps:
Step a) brings into operation from system, every 2 seconds, takes q in the time window1And q2Local maximum be
The method that step b) estimates (Parzen window estimation) using the gloomy window of pa, by eachIt fits
The probability density curve for the vehicle noise energy that systematic observation obtains;
Step c) determines the corresponding energy spectrum size of second wave crest in the probability density curve, is prospect threshold alpha;
It determines energy spectrum size corresponding to the trough between first and second wave crests, is background threshold β.
The energy spectrum of two coarseness detection zones is normalized in step 402);
Step 403) judges whether vehicle passes in and out coarseness detection zone;
WhenWhen, judgement has vehicle to enter first coarseness detection zone;WhenWhen, judgement have vehicle into
Enter second coarseness detection zone;WhenAndWhen, judge that no vehicle enters coarseness detection zone;Wherein
γ is first threshold;In the present embodiment, γ=0.5.
Step 4) starts vehicle count module, calculates vehicle flowrate;
When having sequence of cars after two coarseness detection zones in a lane, add for the vehicle flowrate counting in the lane
1。
Step 5) starts lane occupancy ratio computing module, calculates lane occupancy ratio;
Lane occupancy ratio RtRefer to that vehicle passes through single shared by the cumulative time of coarseness detection zone in unit observation time
The percentage of position observation time:
Wherein tiIt is i-th vehicle by the time used in coarseness detection zone, N is in observation time by detection section
Vehicle number, T is observation time;Each vehicle can be recorded while statistical vehicle flowrate to be spent by coarseness detection zone
Time ti, for calculating Rt。
Step 6) starts speed estimation module, estimates speed;
Enter the time difference of two coarseness detection zones by vehicle to estimate speed v:
Wherein t1And t2Respectively vehicle enter a lane two coarseness detection zones time, h be the vehicle
The distance in the lane is set in monitoring;The height that h can be installed by lane position, lane width and monitoring device determines.
Step 7) starts vehicle classification module, and the classification of large and small vehicle is carried out to vehicle;
The calculation method of Vehicle length are as follows:
L=(tout-tin)v
Wherein, tinAnd toutRespectively vehicle enters and is driven out to time of a coarseness detection zone;
As l >=l0When, judge to pass through vehicle as large car, is otherwise compact car;Wherein l0For second threshold, unite for experience
The preferred value of meter or the threshold value selected automatically by machine learning method.
Claims (7)
1. a kind of device for monitoring vehicle based on audio, which is characterized in that described device includes:
Microphone array module: for acquiring and handling the noise signal of vehicle sending;Each array element acquisition in microphone array
Vehicle noise time-domain signal after analog-to-digital conversion, form the time domain data of horizontal and vertical subarray;To time domain data into
Row multichannel Short Time Fourier Transform, obtains its frequency domain data;Sub-band and time interval is selected to calculate horizontal and vertical son
The correlation matrix C of array;
The microphone array takes cross array structure;Horizontal and vertical subarray all uses equidistant word array knot
Structure, and the array number of two subarrays is different with array element spacing;The low-limit frequency f for the signal that the microphone array can be handledmin
More than or equal to 7kHz;And laterally and longitudinally the array element spacing of subarray should meet dh<λmin/ 2, dv<λmin, wherein λminFor signal
Minimum wavelength;Thereby determine that microphone array can handle the maximum frequency f of signalmax, lateral subarray array number MhWith
Array element spacing dh, longitudinal subarray array number MvWith array element spacing dv;
Lane position and width computing module: for calculating position and the width in each lane;
Coarseness detection zone energy spectrum computing module: for constructing two coarseness detection zones, and benefit on each lane
The energy spectrum in two coarseness detection zones is calculated with correlation matrix C;
Automatic growth control module: for calculating prospect threshold alpha and background threshold β, to the energy in two coarseness detection zones
Amount spectrum is normalized, and judges whether vehicle passes in and out two coarseness detection zones;
Vehicle count module: the vehicle number passed through for counting each lane;
Lane occupancy ratio computing module: for calculating the occupation rate in each lane;
Speed estimation module: for estimating the speed of vehicle;
Vehicle classification module: size type used for vehicles is classified.
2. a kind of vehicle monitoring method based on audio, real based on the device for monitoring vehicle described in claim 1 based on audio
It is existing, which comprises
Step 1) starts microphone array module, and the vehicle noise time-domain signal of each array element acquisition passes through mould in microphone array
After number conversion, the time domain data of horizontal and vertical subarray is formed;Multichannel Short Time Fourier Transform is carried out to time domain data, is obtained
To its frequency domain data;Selection sub-band and time interval calculate the correlation matrix C of horizontal and vertical subarray;
Step 2) starts lane position and width computing module, calculates position and the width in each lane;
Position and the width for obtaining the lane of the monitored highway of the device for monitoring vehicle are high using the installation of device for monitoring vehicle
Degree, at elevation angle theta=0 °, calculates the corresponding azimuthal range in each lane:
Step 3) starts coarseness detection zone energy spectrum computing module, and two coarseness detection zones are constructed on each lane
Domain, and the energy spectrum q in two coarseness detection zones is calculated using correlation matrix C1And q2;
Step 4) starts automatic growth control module, calculates prospect threshold alpha and background threshold β;And to two coarseness detection zones
Energy spectrum on domain is normalized;Judge whether vehicle passes in and out two coarseness detection zones;
Step 5) starts vehicle count module, calculates vehicle flowrate;
When having sequence of cars after two coarseness detection zones in a lane, for the vehicle flowrate in the lane, count is incremented;
Step 6) starts lane occupancy ratio computing module, calculates lane occupancy ratio;
Step 7) starts speed estimation module, estimates speed;
Step 8) starts vehicle classification module, and the classification of large and small vehicle is carried out to vehicle.
3. the vehicle monitoring method according to claim 2 based on audio, which is characterized in that the step 1) is further wrapped
It includes:
The time-domain signal of each array element acquisition vehicle noise of step 101) microphone array, obtains transverse direction after analog-to-digital conversion
The time domain data x of subarray and longitudinal subarrayh(t) and xv(t);
Step 102) is to time domain data xh(t) and xv(t) multichannel Short Time Fourier Transform is carried out, frequency domain data x is obtainedh(f,τ)
And xv(f, τ), wherein f is frequency range number, and τ is the serial number of data frame;
Step 103) selects sub-band [f1,f2] and the corresponding correlation matrix of the horizontal and vertical subarray of time interval Δ τ calculating
C:
Wherein, fmin<f1<f2<fmax, subscript H expression conjugate transposition.
4. the vehicle monitoring method according to claim 3 based on audio, which is characterized in that the step 4) is further wrapped
It includes:
Step 401) calculates prospect threshold alpha and background threshold β;
The energy spectrum of two coarseness detection zones is normalized in step 402);
Step 403) judges whether vehicle passes in and out two coarseness detection zones;
WhenWhen, judgement has vehicle to enter first coarseness detection zone;WhenWhen, judgement has vehicle to enter the
Two coarseness detection zones;WhenAndWhen, judge that no vehicle enters coarseness detection zone;Wherein γ is
First threshold.
5. the vehicle monitoring method according to claim 4 based on audio, which is characterized in that the step 401) is further
Include:
Step a) brings into operation from system, every 2 seconds, takes q in the time window1And q2Local maximum be
The method that step b) is estimated using the gloomy window of pa, by eachFit the general of the vehicle noise energy that systematic observation obtains
Rate density curve;
Step c) determines the corresponding energy spectrum size of second wave crest in the probability density curve, is prospect threshold alpha;It determines
Energy spectrum size corresponding to trough between first and second wave crests is background threshold β.
6. the vehicle monitoring method according to claim 4 based on audio, which is characterized in that the estimation vehicle of the step 7)
The specific implementation process of speed are as follows:
Two coarseness detection zones are constructed in every lane, and the elevation angle in the two regions is respectively θ=- δ and θ=δ, wherein δ
=5 °, the time difference of two coarseness detection zones is entered by vehicle to estimate speed v:
Wherein t1And t2Respectively vehicle enter a lane two coarseness detection zones time, h be the vehicle monitoring
Distance of the device to the lane;H is determined by the height of lane position, lane width and monitoring device installation.
7. the vehicle monitoring method according to claim 6 based on audio, which is characterized in that vehicle in the step 8)
Carry out the specific implementation process of the classification of large and small vehicle are as follows:
The calculation method of Vehicle length are as follows:
L=(tout-tin)v
Wherein, tinAnd toutRespectively vehicle enters and is driven out to time of a coarseness detection zone;
As l >=l0When, judge to pass through vehicle as large car;It otherwise is compact car, wherein l0For second threshold.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510073084.6A CN105989710B (en) | 2015-02-11 | 2015-02-11 | A kind of device for monitoring vehicle and method based on audio |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510073084.6A CN105989710B (en) | 2015-02-11 | 2015-02-11 | A kind of device for monitoring vehicle and method based on audio |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105989710A CN105989710A (en) | 2016-10-05 |
CN105989710B true CN105989710B (en) | 2019-01-22 |
Family
ID=57041884
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510073084.6A Active CN105989710B (en) | 2015-02-11 | 2015-02-11 | A kind of device for monitoring vehicle and method based on audio |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105989710B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109410579B (en) * | 2018-11-12 | 2021-07-06 | 广西交通科学研究院有限公司 | Audio detection system and detection method for moving vehicle |
CN110441390B (en) * | 2019-07-18 | 2021-12-07 | 上海大学 | Damage positioning method based on cross array and space-wave number filter |
CN111163571B (en) * | 2019-12-31 | 2022-04-01 | 北京声智科技有限公司 | Street lamp control method, electronic equipment and computer readable storage medium |
CN112532941A (en) * | 2020-11-30 | 2021-03-19 | 南京中科声势智能科技有限公司 | Vehicle source intensity monitoring method and device, electronic equipment and storage medium |
CN113313951B (en) * | 2021-07-30 | 2021-11-16 | 宁波均联智行科技股份有限公司 | Traffic control method and device based on environmental noise |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5798983A (en) * | 1997-05-22 | 1998-08-25 | Kuhn; John Patrick | Acoustic sensor system for vehicle detection and multi-lane highway monitoring |
US6195608B1 (en) * | 1993-05-28 | 2001-02-27 | Lucent Technologies Inc. | Acoustic highway monitor |
CN102378918A (en) * | 2009-01-20 | 2012-03-14 | 索尼特技术公司 | Acoustic position-determination system |
CN102682765A (en) * | 2012-04-27 | 2012-09-19 | 中咨泰克交通工程集团有限公司 | Expressway audio vehicle detection device and method thereof |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7349849B2 (en) * | 2001-08-08 | 2008-03-25 | Apple, Inc. | Spacing for microphone elements |
US7171008B2 (en) * | 2002-02-05 | 2007-01-30 | Mh Acoustics, Llc | Reducing noise in audio systems |
-
2015
- 2015-02-11 CN CN201510073084.6A patent/CN105989710B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6195608B1 (en) * | 1993-05-28 | 2001-02-27 | Lucent Technologies Inc. | Acoustic highway monitor |
US5798983A (en) * | 1997-05-22 | 1998-08-25 | Kuhn; John Patrick | Acoustic sensor system for vehicle detection and multi-lane highway monitoring |
CN102378918A (en) * | 2009-01-20 | 2012-03-14 | 索尼特技术公司 | Acoustic position-determination system |
CN102682765A (en) * | 2012-04-27 | 2012-09-19 | 中咨泰克交通工程集团有限公司 | Expressway audio vehicle detection device and method thereof |
Also Published As
Publication number | Publication date |
---|---|
CN105989710A (en) | 2016-10-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN105989710B (en) | A kind of device for monitoring vehicle and method based on audio | |
Odat et al. | Vehicle classification and speed estimation using combined passive infrared/ultrasonic sensors | |
CN108510797B (en) | Forewarning System of Freeway and method based on radar detecting | |
CN105261215B (en) | Intelligent transportation behavior cognitive method and system based on microwave | |
Kwong et al. | Arterial travel time estimation based on vehicle re-identification using wireless magnetic sensors | |
US5798983A (en) | Acoustic sensor system for vehicle detection and multi-lane highway monitoring | |
US11030896B2 (en) | Real-time traffic information collection | |
JPH06194443A (en) | Computing system for investigation parameter of traffic of one or more vehicles | |
JPH09512100A (en) | Traffic surveillance for automatic vehicle incident detection | |
CN111582160B (en) | Vehicle track extraction method and device based on distributed optical fiber sensing system | |
CN101751782A (en) | Crossroad traffic event automatic detection system based on multi-source information fusion | |
CN208422108U (en) | Transport investigation device based on millimetre-wave radar and laser radar | |
CN202352081U (en) | Intelligent traffic light control system based on wireless sensor network | |
CN113791410B (en) | Road environment comprehensive cognition method based on multi-sensor information fusion | |
CN114858214B (en) | Urban road performance monitoring system | |
CN101131796A (en) | Road traffic parameter checking device and method thereof | |
CN114003849A (en) | Multi-lane non-contact type axle number automatic calculation method and system | |
Wu et al. | A novel skateboarder-related near-crash identification method with roadside LiDAR data | |
CN202995036U (en) | Radar echo characteristic based system for detecting and alarming railway roadblocks | |
CN114999113A (en) | Real-time prediction method for icing state of highway pavement | |
CN105989709B (en) | A kind of road driveway dynamic positioning method based on audio | |
CN106803348A (en) | A kind of overpass vehicle travel time Forecasting Approach for Short-term based on bayonet socket | |
Lücken et al. | Density-based statistical clustering: enabling sidefire ultrasonic traffic sensing in smart cities | |
Gunawan et al. | Optimal averaging time for predicting traffic velocity using floating car data technique for advanced traveler information system | |
CN116165655A (en) | Global vehicle track tracking system based on millimeter wave radar group |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |