The content of the invention
In order to solve the above problems, the invention provides a kind of moving target detecting method based on frequency modulated continuous wave radar
And system.A kind of moving target detecting method and system based on frequency modulated continuous wave radar provided by the invention, are connected using frequency modulation
The method that continuous ripple radar is detected to moving vehicle target, more specifically apply to the tree that random swing with the wind be present
In the scene of leaf interference, echo data is handled and estimates the detection method of moving vehicle speed and distance.
The technical solution adopted by the present invention is as follows:
A kind of moving target detecting method based on frequency modulated continuous wave radar, comprises the following steps:
Obtain by frequency modulated continuous wave radar continuously scan the foreground data I of moving target in the detection range repeatedly obtained with
And the frequency modulation section echo-signal frequency data after the difference frequency that is formed of background data I;
Frequency spectrum data corresponding to the frequency modulation section echo-signal frequency data after difference frequency is obtained using Fourier transformation;
Gauss hybrid models are established based on echo spectrum data, by foreground data II corresponding in the frequency spectrum data
Separated with background data II;
The movement velocity and direction of motion data of moving target are obtained using Hough transformation based on the foreground data II.
A kind of above-mentioned moving target detecting method based on frequency modulated continuous wave radar, wherein, it is described to be based on echo spectrum
Data, which establish gauss hybrid models, to be included:
According to the frequency spectrum data, a gauss hybrid models are established for the frequency point data that each frequency is j Hz,
After removing the gauss hybrid models repeated, K gauss hybrid models are obtained.
A kind of above-mentioned moving target detecting method based on frequency modulated continuous wave radar, wherein, it is described by the spectrum number
In corresponding foreground data II separated with background data II including:
Range value g of the frequency that each frequency is j Hz within its corresponding cyclejThe probability of appearance is:
Wherein, wK, jThe weight of the frequency that frequency is j Hz in k-th of gauss hybrid models is represented,
Represent Gaussian probability-density function, μK, jRepresent gjAverage,Represent gjVariance;
B gauss hybrid models are chosen from K gauss hybrid models meets following condition as the correspondence of background data II
Distribution:
Wherein, Td is the threshold value of setting.
A kind of above-mentioned moving target detecting method based on frequency modulated continuous wave radar, wherein, it is described to be based on echo spectrum
Data establish gauss hybrid models, and foreground data II corresponding in the frequency spectrum data is separated with background data II
Also include gauss hybrid models to update, the gauss hybrid models renewal includes:
The frequency spectrum data in new cycle is read, by the range value of each frequency therein and existing K Gauss
Mixed model is compared, and judges whether to meet Mahalanobis distances;
If it is, the relevant parameter of K gauss hybrid models of renewal, the relevant parameter include μK, j、
If not, increase new gauss hybrid models.
A kind of above-mentioned moving target detecting method based on frequency modulated continuous wave radar, wherein, it is described to be based on the prospect
Data II obtain the movement velocity of moving target with also including after direction of motion data using Hough transformation:
Judge whether to continue to detect moving target,
If it is, extremely described acquisition continuously scans moving target in the detection range repeatedly obtained by frequency modulated continuous wave radar
Foreground data I and the difference frequency that is formed of background data I after frequency modulation section echo-signal frequency data;
If not, detection of end.
A kind of moving object detection system based on frequency modulated continuous wave radar, including:
Scan module, moving target in the detection range repeatedly obtained is continuously scanned by frequency modulated continuous wave radar for obtaining
Foreground data I and the difference frequency that is formed of background data I after frequency modulation section echo-signal frequency data;
Frequency domain modular converter, it is corresponding for obtaining the frequency modulation section echo-signal frequency data after difference frequency using Fourier transformation
Frequency spectrum data;
Data separating module, for establishing gauss hybrid models based on echo spectrum data, by the frequency spectrum data with
Its corresponding foreground data II is separated with background data II;
Motion capture module, for obtaining the motion of moving target using Hough transformation based on the foreground data II
Speed and direction of motion data.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, the data separating module
Unit is established including model, the model is established subelement and is used for according to the frequency spectrum data, is j Hz for each frequency
Frequency point data establish a gauss hybrid models, after removing the gauss hybrid models repeated, obtain K Gaussian Mixture mould
Type.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, the data separating module
Including also including data separating unit, the data separating unit is used for foreground data corresponding in the frequency spectrum data
II is separated with background data II, and the data separating unit includes:
Probability obtains subelement, and the range value g in cycle is corresponded at it for calculating the frequency that each frequency is j Hzj
The probability of appearance is:
Wherein, wK, jThe weight of the frequency that frequency is j Hz in k-th of gauss hybrid models is represented,
Represent Gaussian probability-density function, μK, jRepresent gjAverage,Represent gjVariance;
Background data II separates subelement, for chosen from K gauss hybrid models B gauss hybrid models satisfaction with
Lower condition is distributed as corresponding to background data II:
Wherein, Td is the threshold value of setting.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, the data separating module
Also include model modification unit, the model modification unit updates for gauss hybrid models, and the model modification unit includes:
Judgment sub-unit, for reading the frequency spectrum data in new cycle, by the range value of each frequency therein
Compared with existing K gauss hybrid models, judge whether to meet Mahalanobis distances;
Data update subelement, for when judging structure to be, updating K gauss hybrid models based on judgment sub-unit
Relevant parameter, the relevant parameter includes μK, j、
Model modification subelement, for the decision mechanism based on judgment sub-unit for it is no when, increase new Gaussian Mixture mould
Type.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, in addition to continue detection and sentence
Disconnected module, for judging whether to continue to detect moving target, continue to detect when judging structure to be of judge module when described, extremely
The scan module;Continue to detect when judging structure to be no of judge module, the detecting system detection of end when described.
The present invention carries out frequency domain conversion so as to carry out moving target based on the frequency modulation section echo-signal frequency data after difference frequency
Foreground data II and background data II separation, the frequency modulation section echo-signal frequency data after difference frequency postpone to produce by target range
Raw frequency displacement and Doppler frequency two parts synthesizes to obtain.Frequency is in each frequency modulation only caused by the range delay of moving target
What the section cycle changed with motion state, and the frequency displacement of static target is constant, swings echo frequency displacement caused by leaf in each week
Approximate Gaussian distribution on phase frequency spectrum, therefore, the spectrum amplitude figure in each cycle is modeled using mixed Gaussian mixed model, it is static
The echo spectrum of target and leaf can be judged as background dot, and the frequency spectrum point of moving target can be extracted as prospect, and
Estimate kinematic parameter.
The present invention can reach following technique effect:First, in the complex scene swung with the wind for leaf be present, using tune
Frequency continuous wave radar detects vehicle target speed and position.2nd, the design of radar emission waveform can be simplified, it is only necessary to it is single just
Frequency sweep or negative swept continuous wave can be completed to detect, it is not necessary to the complicated modulation waveform such as triangular wave.3rd, robustness is good, accurately
Degree is high, and Parameter adjustable, suitable for varying environment and the detecting and tracking of different target, clutter pair especially caused by wind leaf
The influence very little of the algorithm.
Embodiment
As shown in figure 1, the characteristics of moving target is monitored based on frequency modulated continuous wave radar in the embodiment of the present invention, there is provided one
Moving target detecting method of the kind based on frequency modulated continuous wave radar, comprises the following steps:
S1, obtain the foreground data that moving target in the detection range repeatedly obtained is continuously scanned by frequency modulated continuous wave radar
I and the difference frequency that is formed of background data I after frequency modulation section echo-signal frequency data;
Frequency modulated continuous wave radar run-down, that is, within a scan period, obtain the motion mesh in scanning range
Corresponding data under one state of target, for moving target, it is necessary to which memory Multiple-Scan, every time scanning obtain a motion shape
State, the data of Multiple-Scan obtain the data of moving target in motion process at different moments.
Assuming that frequency modulated continuous wave radar continuously scans M times, you can obtains the frequency modulation section echo-signal after the difference frequency in M cycle
Frequency data.
It is understood that the foreground data by both having included moving target in the data of frequency modulated continuous wave radar scanning,
Including background data, that is to say, that the frequency modulation section echo-signal frequency data after the difference frequency are the M cycles by moving target
What foreground data I and background data I constructed.
S2, frequency spectrum data corresponding to the frequency modulation section echo-signal frequency data after difference frequency is obtained using Fourier transformation;
Frequency modulation section echo-signal frequency data after the difference frequency in each cycle are obtained into corresponding frequency using Fourier transformation
Modal data, then the frequency spectrum data in each cycle includes R frequency, and the frequency of each frequency is that j Hz are width corresponding to the frequency
Angle value gj。
S3, gauss hybrid models are established based on echo spectrum data, by prospect number corresponding in the frequency spectrum data
Separated according to II with background data II;Specifically comprise the following steps:
S31, according to the frequency spectrum data, a Gaussian Mixture mould is established for the frequency point data that each frequency is j Hz
Type, after removing the gauss hybrid models repeated, obtain K gauss hybrid models.
For each frequency, it has M gj, a gauss hybrid models are established, i.e., if R frequency, just establish R
Individual gauss hybrid models, still, there are the gauss hybrid models repeated in R gauss hybrid models, therefore, remove and repeat
After existing gauss hybrid models, K gauss hybrid models are obtained.
S32, foreground data II corresponding in the frequency spectrum data is separated with background data II including:
Range value g of the frequency that each frequency is j Hz within its corresponding cyclejThe probability of appearance is:
Wherein, wK, jThe weight of the frequency that frequency is j Hz in k-th of gauss hybrid models is represented,
Represent Gaussian probability-density function, μK, jRepresent gjAverage,Represent gjVariance;
B gauss hybrid models are chosen from K gauss hybrid models meets following condition as the correspondence of background data II
Distribution:
Wherein, Td is the threshold value of setting.
For the fixed buildings target in detection scene, in each peri odic spectrum data in corresponding range value relatively not
Become, variance is smaller, therefore matches the distribution characteristics of background quickly.The target of the irregular movement such as leaf, each during for wind
Although the position and amplitude that the spectrogram in cycle occurs are mixed and disorderly, dynamic range is little, for static object
Variance is larger, and the feature of background branch can be just met after the study in multiple cycles.But moving target pushes away over time
Move, frequency peak is constantly offset on the spectrogram in each cycle, and variance changes greatly, therefore will not meetDecision condition.To sum up, meetJudgement bar
The frequency spectrum data of part is the background data II of moving target;It will not meetJudgement bar
The frequency spectrum data of part is the foreground data II of moving target.
It is understood that the foreground data II in step S3 is different from the foreground data in step S1 with background data II
I and background data I, in step s3, foreground data II belongs to frequency domain data with background data II, and the frequency spectrum data includes
The foreground data II of moving target, also including background data II, that is to say, that the frequency spectrum data is the M week by moving target
The foreground data II of phase constructs with background data II.
S4, the movement velocity and direction of motion number of moving target are obtained using Hough transformation based on the foreground data II
According to.The spectrogram data combination in multiple cycles of mixed Gauss model will be established, within each cycle, removes range value, judges
The frequency values for whether having corresponding frequency occur, and establish the corresponding relation data in frequency and cycle, so as to obtain each cycle inspection
The curves of kinetic feature of the moving target measured, using the corresponding curve movement of Hough change detection and kinematic parameter is estimated,
The kinematic parameter includes movement velocity, the direction of motion.
In one embodiment, above-mentioned a kind of moving target detecting method based on frequency modulated continuous wave radar, the step
S3 also includes step S33 gauss hybrid models and updated, and the step S33 includes:
S331, the frequency spectrum data in new cycle is read, by the range value of each frequency therein and existing K
Gauss hybrid models are compared, and judge whether to meet Mahalanobis distances;If it is, to step S332;If not, extremely
Step S333.
S332, if it is, the relevant parameter of K gauss hybrid models of renewal, the relevant parameter include μK, j、
Learning rate etc..
S333, if not, increasing new gauss hybrid models, new gauss hybrid models are with the frequency spectrum data of current period
In average and variance as model of average and variance, weight is empirical value.
In one embodiment, above-mentioned a kind of moving target detecting method based on frequency modulated continuous wave radar, wherein, also wrap
Step S5 is included, judges whether to continue to detect moving target, if it is, to step S1;If not, detection of end.
A kind of moving object detection system based on frequency modulated continuous wave radar, including:
Scan module, moving target in the detection range repeatedly obtained is continuously scanned by frequency modulated continuous wave radar for obtaining
Foreground data I and the difference frequency that is formed of background data I after frequency modulation section echo-signal frequency data;
Frequency domain modular converter, it is corresponding for obtaining the frequency modulation section echo-signal frequency data after difference frequency using Fourier transformation
Frequency spectrum data;
Data separating module, for establishing gauss hybrid models based on echo spectrum data, by the frequency spectrum data with
Its corresponding foreground data II is separated with background data II;
Motion capture module, for obtaining the motion of moving target using Hough transformation based on the foreground data II
Speed and direction of motion data.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, the data separating module
Unit is established including model, the model is established subelement and is used for according to the frequency spectrum data, is j Hz for each frequency
Frequency point data establish a gauss hybrid models, after removing the gauss hybrid models repeated, obtain K Gaussian Mixture mould
Type.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, the data separating module
Including also including data separating unit, the data separating unit is used for foreground data corresponding in the frequency spectrum data
II is separated with background data II, and the data separating unit includes:
Probability obtains subelement, and the range value g in cycle is corresponded at it for calculating the frequency that each frequency is j Hzj
The probability of appearance is:
Wherein, wK, jThe weight of the frequency that frequency is j Hz in k-th of gauss hybrid models is represented,
Represent Gaussian probability-density function, μK, jRepresent gjAverage,Represent gjVariance;
Background data II separates subelement, for chosen from K gauss hybrid models B gauss hybrid models satisfaction with
Lower condition is distributed as corresponding to background data II:
Wherein, Td is the threshold value of setting.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, the data separating module
Also include model modification unit, the model modification unit updates for gauss hybrid models, and the model modification unit includes:
Judgment sub-unit, for reading the frequency spectrum data in new cycle, by the range value of each frequency therein
Compared with existing K gauss hybrid models, judge whether to meet Mahalanobis distances;
Data update subelement, for when judging structure to be, updating K gauss hybrid models based on judgment sub-unit
Relevant parameter, the relevant parameter includes μK, j、
Model modification subelement, for the decision mechanism based on judgment sub-unit for it is no when, increase new Gaussian Mixture mould
Type.
A kind of above-mentioned moving object detection system based on frequency modulated continuous wave radar, wherein, in addition to continue detection and sentence
Disconnected module, for judging whether to continue to detect moving target, continue to detect when judging structure to be of judge module when described, extremely
The scan module;Continue to detect when judging structure to be no of judge module, the detecting system detection of end when described.
Fig. 2 be experiment in frequency modulated continuous wave radar transmitted waveform, carrier frequency 24GHz.The present invention is not related to the measurement of angle,
Therefore the echo data of a reception antenna has only been used.There was only a moving vehicle target in scene when detecting, and set
The influence of the random swing of leaf.The clutter spectrum caused by leaf swing masks the frequency spectrum of moving target, traditional frequency spectrum
Matching algorithm can not obtain the motion conditions of target.Wherein Fig. 3 be frequency modulated continuous wave radar signal processing module to it is current when
The result of motion measured target parameter Estimation is carved, sector region represents the detectable scope of radar, and the point in region represents detection
The moving-target gone out, parameter include distance, angle and the speed of target.Only existed in scene under a moving vehicle target conditions,
The module detected up to three false-alarm targets.Fig. 4 is institute's rhythmic positive frequency sweep section spectrum number of the invention to having gathered
According to carrying out gauss hybrid models modeling, and extract with Hough transformation the image of vehicle target movement locus on frequency spectrum.Horizontal seat
Mark represents number of detection cycles, and ordinate represents spectral sample position, finds out that vehicle travels away from radar by slope is positive and negative, calculates
It is 21.5Km/h to go out target in the speed v of current period, is 59.1m apart from radar R.
When detection scene includes two moving vehicle targets.Under identical scene, wherein Fig. 5 is frequency modulated continuous wave radar
There are four false-alarm targets to the result of current time moving target parameter detecting in processing module.
Fig. 6 is that the inventive method carries out gauss hybrid models modeling to positive frequency sweep partial frequency spectrum of all cycles, and uses Hough
Conversion extracts the image of vehicle target movement locus on frequency spectrum.Same abscissa represents number of detection cycles, and ordinate represents
Spectral sample position, it can be seen that two straight slopes are identical, i.e. two vehicle targets are synchronized in the same direction and travelled close to radar, but
It is that this partial straight lines all part clutter point occurs in extraction process in end position, causes to have extracted two slopes near
As straight line, this is due to now nearer apart from radar, close to the detectable minimum range of the radar, therefore the meeting at the point of end
There is deviation.According to the parameter of two straight lines, the speed v that two targets are calculated is 19.1Km/h, wherein a car distance
Distance by radar R is 7.3m, and another car sails out of radar before eight cycles, beyond radar observation region.
The present invention carries out vehicle target detection using frequency modulated continuous wave radar.The transmission signal of frequency modulated continuous wave radar is to sweep
Frequency signal, can be continuous positive frequency sweep or negative swept-frequency signal.Swept-frequency signal can use triangular signal, sawtooth signal etc.
Form.The detection of the invention for only needing sawtooth signal to complete moving target.It should be noted that used due to experiment
It is fmcw radar, transmission signal can only use the triangular signal of fixed cycle, but only with each frequency sweep week during calculating
The positive frequency sweep segment signal of phase replaces sawtooth signal, meets the principle of the present invention.
In addition, the part of the application can be applied to computer program product, such as computer program instructions, when its quilt
When computer performs, by the operation of the computer, it can call or provide according to the present processes and/or technical scheme.
And the programmed instruction of the present processes is called, it is possibly stored in fixed or moveable recording medium, and/or pass through
Broadcast or the data flow in other signal bearing medias and be transmitted, and/or be stored according to described program instruction operation
In the working storage of computer equipment.Here, including a device according to one embodiment of the application, the device includes using
Memory in storage computer program instructions and processor for execute program instructions, wherein, when the computer program refers to
When order is by the computing device, method and/or skill of the plant running based on foregoing multiple embodiments according to the application are triggered
Art scheme.
It is obvious to a person skilled in the art that the application is not limited to the details of above-mentioned one exemplary embodiment, Er Qie
In the case of without departing substantially from spirit herein or essential characteristic, the application can be realized in other specific forms.Therefore, no matter
From the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, scope of the present application is by appended power
Profit requires rather than described above limits, it is intended that all in the implication and scope of the equivalency of claim by falling
Change is included in the application.Any reference in claim should not be considered as to the involved claim of limitation.This
Outside, it is clear that the word of " comprising " one is not excluded for other units or step, and odd number is not excluded for plural number.That is stated in device claim is multiple
Unit or device can also be realized by a unit or device by software or hardware.The first, the second grade word is used for table
Show title, and be not offered as any specific order.
Certainly, it is obvious to a person skilled in the art that the application is not limited to the details of above-mentioned one exemplary embodiment, and
And in the case of without departing substantially from spirit herein or essential characteristic, the application can be realized in other specific forms.Therefore,
No matter from the point of view of which point, embodiment all should be regarded as exemplary, and be nonrestrictive, scope of the present application is by institute
Attached claim rather than described above limit, it is intended that will fall in the implication and scope of the equivalency of claim
All changes are included in the application.Any reference in claim should not be considered as to the involved right of limitation will
Ask.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all essences in the present invention
God any modification, equivalent substitution and improvements made etc., should be included in the scope of the protection with principle.