CN101349754B - Automatic recognition method of road driveway based on millimeter wave traffic radar - Google Patents

Abstract

A road driveway automatic recognition method based on millimeter wave traffic radar adopts adaptive threshold value to check present road structure information, adopts the adaptive threshold value to check the effective vehicle power spectrum containing a radar maximum echo power, normalizes the effective vehicle power spectrum, enters into a power accumulation process, adopts a window function method after collecting enough data to smoothen the normalized accumulation power spectrum to filter out the high frequency part to improve its morphology, recognizes the valley of the normalized accumulation power spectrum as the boundary of present road driveway, and utilizes self-learning method to real-timely refresh the boundary value of the road driveway. The invention has the advantages that the road driveway automatic recognition method can automatically recognize road driveways in short time, can automatically and real-timely refresh the driveway recognition according to the traffic mode and weather, can accurately and real-timely recognize road driveways and can avoid manual operation.

Description

Automatic recognition method of road driveway based on millimeter wave traffic radar

Technical field

The present invention relates to the traffic information collection field in the intelligent transportation, be specifically related to a kind of automatic recognition method of road driveway based on millimeter wave traffic radar.

Background technology

Traffic information acquisition system is the indispensable element of intelligent transportation system.Traffic information acquisition system depends on traffic information sensor, and the ability that sensor obtains traffic data in real time exactly all is vital to traffic control, public safety and traffic programme.Modern city is along with development economic and society becomes more crowded on the contrary, and traffic problems have caused widely to be paid close attention to.The quickening of rhythm of life makes people expect to have more good traffic more.

Many kinds of traffic information collection equipment have obtained utilization widely, as the buried type coil checker applicating history of five more than ten years have been arranged, and have used so far.But this type of buried type transport information detecting device all has higher crash rate usually, and higher installation and maintenance cost is arranged.More bad is that the I﹠M of this type of buried type transport information detecting device can cause pavement destruction and track to be closed.In recent decades, the transport information checkout equipment of the non-buried type of many kinds has obtained using widely, and this is comprising acoustics and optical traffic information sensor.But these traffic information sensors have significant disadvantages.At first, these sensors can only detect the wall scroll runway, and powerless to the traffic that detects the whole piece highway; The second, this class sensor is very responsive to outside environmental change usually, such as Changes in weather, changes or the like round the clock.

Appearing to a certain extent of millimeter wave traffic radar solves these problems, the variable power of the different frequency section by analyzing different tracks correspondence, and millimeter wave traffic radar can be realized the function that the multilane transport information detects.The advantage that millimeter wave traffic radar has many other transport information detecting devices to hardly match as all weather operations, does not have the blind area of detection, and low installation and maintenance cost does not cause damage or the like to highway pavement.These good characteristics of millimeter wave traffic radar make it become the good substitute of buried type coil and acousto-optic transport information detecting device.

The whether accurate detection performance that has directly influenced millimeter wave traffic radar of road driveway identification, the inefficacy of radar is discerned even can be caused to the road driveway of mistake.In actual applications, need manual adjustments and calibration millimeter wave traffic radar to finish the work of lane identification usually.But, all need manually to readjust and calibrate radar along with the variation of highway pattern and reinstalling of radar.In addition, in the different periods of every day, the variation of weather as day and night, sleet covering etc., can cause the lane shift of millimeter wave traffic radar equally.In this case, though radar operate as normal still, detected transport information is no longer accurate, the communications policy that this may lead to errors.This shows, road driveway automatically identification be important also be necessary.

China Patent No. 1912949A, patent name is a kind of radar road identification method for detecting vehicle flow, discloses a kind of hardware system and traffic information detection method of millimeter wave traffic flow detection radar.Lane identification in this method still relies on manually and regulates.

United States Patent (USP) 7091901B2, patent name is System and Method forIdentification of Traffic Lane Positions, a kind of dynamic auto recognition methods of road driveway of millimeter wave traffic radar is disclosed, the probability density function that this method utilizes the vehicle power spectrum peak position to occur is realized the dynamic auto identification of road driveway, this method needs the long time to the identification in track, be difficult to the requirement that reaches real-time, can't in time detect the track that causes by unexpected precipitation and change.In addition, there are a plurality of minimal values in the probability density function that this method obtains, and the minimal value position on road driveway border has been represented in very difficult identification.

Summary of the invention

In order to solve above technical matters, the invention provides a kind of automatic recognition method of road driveway based on millimeter wave traffic radar, this recognition methods comprises the steps:

Step 1: utilize adaptive threshold to detect current highway structure information;

Step 2: identification comprises effective vehicle power spectrum of the maximum echo power of radar;

Step 3: effectively vehicle normalized power accumulation;

Step 4: level and smooth normalization cumulative power spectrum;

Step 5: discern current road driveway border;

Step 6: utilize the self study process to realize road driveway identification.

Described step 1 further comprises: the adaptive threshold setting; Highway structure identification.

Described adaptive threshold setting at each frequency place, will be imported highway echo power P _RTime series sort according to power magnitude, obtain orderly echo power sequence P _R0s, adaptive threshold T at Frequency point k place _AdpCan be expressed as:

$T_{\mathrm{adp}}\left(k\right)=\frac{o}{b}\underset{i=\mathrm{aM}}{\overset{(a+b)M}{\mathrm{\Σ}}}{P}_{\mathrm{Ros}}(k,i)$ k∈[K ₁，K ₂]，

In the formula, a is normalization valid data starting points, and b is normalization valid data capacity, and M is a total data capacity, and o is the threshold value multiplier, [K ₁, K ₂] for having comprised the frequency band of track full detail, i is a Frequency point.

Described highway structure identification is by detecting adaptive threshold T _AdpAmplitude size realize that highway structure is meant the highway facilities that the runway of difference in functionality is separated, as guard rail, greenbelt.If T _AdpAmplitude satisfy following condition, then can think to have this type of highway structure on the highway:

$T_{\mathrm{adp}}>c\·\underset{k=K_1}{\overset{K_2}{\mathrm{\Σ}}}{T}_{\mathrm{adp}}\left(k\right).$

In the formula, c is the highway structure critical parameter, $c\⋐\left[\mathrm{2,3}\right].$

Described step 3 further comprises: effectively vehicle power spectrum normalization; The normalized power accumulation.

Following condition is satisfied in the normalization of described effective vehicle power spectrum:

$\left\{\begin{array}{c}{P}_{\mathrm{vnorm}}\left(k\right)=\frac{(P_R\left(k\right)-T_{\mathrm{adp}}\left(k\right))\·g\left(k\right)}{\max \left(T_{\mathrm{adp}}\right)}\\ \underset{k=K_1}{\overset{K_2}{\mathrm{\Σ}}}{P}_{\mathrm{vnorm}}\left(k\right)\≠0\end{array}\right.,$

In the formula, P _VnormBe effective vehicle normalized power spectrum, P _Vnorm(k) be the value of effective vehicle normalized power spectrum, max (T at frequency k place _Adp) expression T _Adp(k) maximal value in, g (k) is effective vehicle discriminant function:

Described normalized power accumulation can be expressed from the next:

$\left\{\begin{array}{cc}\begin{array}{c}{P}_{\ln \mathrm{orm}}^n\left(k\right)=\frac{(n-1)P_{\ln \mathrm{orm}}^{n-1}\left(k\right)+P_{\mathrm{vnorm}}\left(k\right)}{n}\\ P_{\ln \mathrm{orm}}^0\left(k\right)=T_{\mathrm{adp}}\left(k\right)\end{array}& n=1,...,N,\end{array}\right.$

In the formula, P _LnormBe the normalized power cumulative function, n is the sample size that is used for current lane identification.

Described level and smooth normalization cumulative power spectrum realizes that by the smoothing windows function smoothing windows function uses rectangular window:

$w_k\left(n\right)=u(n-k+\frac{W-1}{2})-u(n-k-\frac{W-1}{2}),$

In the formula, u (n) is a step function, and window width W is an odd number;

Normalization cumulative power spectrum after level and smooth can be expressed from the next:

$\stackrel{\‾}{P_{\ln \mathrm{orm}}\left(k\right)}=\frac{\underset{i=K_1}{\overset{K_2}{\mathrm{\Σ}}}(P_{\ln \mathrm{orm}}\left(i\right)\·w_k\left(i\right))}{W}.$

The current road driveway of described identification border, need border and forward and reverse highway border between the identification road driveway, if highway forward or reverse track comprise n bar track, number of boundary is n-1 between the road driveway that needs so to determine, forward or reverse highway number of boundary are 2.

The border is to realize by the minimal value position of the normalization cumulative power spectrum after detecting smoothly between the identification road driveway.Have more vehicle to pass through at the center, track, therefore more power accumulation is arranged, the normalization cumulative power after level and smooth is composed formation power spectrum peak.And the vehicle that lane boundary is passed through is less, can form power spectrum paddy.Power spectrum paddy place is the minimal value place of the normalization cumulative power spectrum after level and smooth, is exactly the position on border between the track.

Identification forward or reverse highway border if there is highway structure information, can utilize highway structure information to finish identification so.If there is no highway structure information can be composed the normalization cumulative power minimal power values in all spectrum paddy as threshold value, and the position that definition power spectrum top and terminal performance number equal this threshold value is forward or reverse highway border.

The described self study process of utilizing realizes road driveway identification, is meant that at each frequency place, the self study process of lane identification can be expressed as:

$L^t\left(k\right)=(1-r)L^{t-1}\left(k\right)+r{L}_0^t\left(k\right).$

L is a t road driveway recognizing site constantly in the formula, L ₀Be current lane identification position, r is a learning rate.

Superior effect of the present invention is:

1) the present invention can finish the automatic identification of road driveway in the short period of time, and can realize lane identification with the variation of travel pattern and weather conditions automatic real-time update;

2) the present invention can accurately discern road driveway in real time, need not any manually-operated.

Description of drawings

Fig. 1 is the process flow diagram that the road driveway of a specific embodiment of the present invention is discerned automatically;

Fig. 2 is that the traffic radar in the specific embodiment of the present invention is installed and the highway environment synoptic diagram;

Fig. 3 is the adaptive threshold testing result in the specific embodiment of the present invention;

Fig. 4 is the effective vehicle power spectrum recognition result that comprises the maximum echo power of radar in the specific embodiment of the present invention;

Fig. 5 is the effective vehicle normalized power accumulation results in the specific embodiment of the present invention;

Fig. 6 is the current road driveway Boundary Recognition result in the specific embodiment of the present invention.

Embodiment

See also shown in the accompanying drawing, the invention will be further described.

According to shown in the process flow diagram of the automatic identification of road driveway of a specific embodiment of the present invention, at first utilize the highway echo power as Fig. 1, calculate adaptive threshold, and the application self-adapting threshold value is carried out current highway structure identification; Identification comprises effective vehicle power spectrum of the maximum echo power of radar then, and carries out the effective vehicle power spectrum accumulation of normalization; After having obtained abundant data, to use the level and smooth normalization useful power spectrum of window function cumulative function, and determine border between current track by the spectrum paddy position of identification normalization cumulative power spectrum, the passing threshold method is determined current highway border; Finish road driveway identification by the self study process at last.

According to shown in installation of the traffic radar in the specific embodiment of the present invention and the highway environment synoptic diagram, on the basis of existed system model machine, we have carried out field test on the road of Changning, Shanghai City as Fig. 2.Millimeter wave traffic radar 201 is installed on the other support 202 of highway, and the beam direction of assurance radar is perpendicular to the highway direction.Millimeter wave traffic radar 201 can detect the transport information of forward highway 204 and reverse highway 206 simultaneously.Highway is two-way eight tracks, and having height between first lane and bicycle lane is the metal protection hurdle 203 of 1.5m.And between forward highway 204 and reverse highway 206, having protecting and greening band 205, this greenbelt is made up of shrub and lawn.In once testing, the first lane of forward highway 204 exists vehicle 207, the Four-Lane Roads to have vehicle 209.

According to shown in the adaptive threshold testing result in the specific embodiment of the present invention, power spectrum peak 303 is to be caused by the metal protection hurdle 203 between first lane and the bicycle lane as Fig. 3.Protecting and greening band 205 has caused power spectrum envelope 305.Because the power spectrum peak 303 that metal protection hurdle 203 produces has surpassed highway structure decision threshold 301, this highway structure can be identified.Because the isolation strip 205 between forward and reverse highway is made up of shrub and lawn, its echo power is less, although there is power envelope 305 to exist, it does not surpass highway structure decision threshold 301, and this highway structure can't be discerned.The maximum value position of the power envelope 303 of structure 203 generations can identify a border 302 of highway by road.

As Fig. 4 according to shown in the recognition result of the effective vehicle power spectrum that comprises the maximum echo power of radar in the specific embodiment of the present invention, the useful power spectrum 409 that the maximal value of the useful power spectrum 407 that the first lane vehicle 207 of forward highway 204 causes causes less than forward highway 204 Four-Lane Road vehicles 209, the useful power that so current effective vehicle power spectrum is forward highway 204 Four-Lane Road vehicles 209 to be caused spectrum 409, this power spectrum is present in interval 401.

As Fig. 5 according to shown in the effective vehicle normalized power accumulation results in the specific embodiment of the present invention, effective vehicle normalized power spectrum 502 of previous moment obtains current effective vehicle normalized power spectrum 501 after having learnt current effective vehicle normalized power spectrum 409.

According to shown in the current road driveway Boundary Recognition result in the specific embodiment of the present invention, can determine that by the minimal value position of the normalization cumulative power spectrum 601 after discerning smoothly the border is 602,603,604 between road driveway as Fig. 6.Because there is not highway structure information in forward highway 204 near the highway border of greenbelt 205, so the minimal power values 603 in normalization cumulative power spectrum 601 all the spectrum paddy after utilizing smoothly is as threshold value 606, the forward highway of determining thus 204 is 605 near the highway border of greenbelt 205.

Claims (10)

1. automatic recognition method of road driveway based on millimeter wave traffic radar, its characteristic is: comprise the steps:

Step 1: utilize adaptive threshold to detect current highway structure information;

Step 2: identification comprises effective vehicle power spectrum of the maximum echo power of radar;

Step 3: effectively vehicle normalized power accumulation;

Step 4: level and smooth normalization cumulative power spectrum;

Step 5: discern current road driveway border;

Step 6: utilize the self study process to realize road driveway identification.

2. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 1, its characteristic is: described step 1 further comprises:

Adaptive threshold is provided with;

Highway structure identification.

3. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 2 is characterized in that:

Described adaptive threshold setting at each frequency place, will be imported highway echo power P _RTime series sort according to power magnitude, obtain orderly echo power sequence P _Ros, adaptive threshold T at Frequency point k place _AdpCan be expressed as:

$T_{\mathrm{adp}}\left(k\right)=\frac{o}{b}\underset{i=\mathrm{aM}}{\overset{(a+b)M}{\mathrm{\Σ}}}{P}_{\mathrm{Ros}}(k,i)$ k∈[K ₁，K ₂]，

In the formula, a is normalization valid data starting points, and b is normalization valid data capacity, and M is a total data capacity, and o is the threshold value multiplier, [K ₁, K ₂] for having comprised the frequency band of track full detail, i is a Frequency point.

4. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 3 is characterized in that:

Described highway structure identification is by detecting adaptive threshold T _AdpAmplitude size realize that highway structure is meant the highway facilities that the runway of difference in functionality is separated, if T _AdpAmplitude satisfy following condition, then can think to have this type of highway structure on the highway:

$T_{\mathrm{adp}}>c\·\underset{k=K_1}{\overset{K_2}{\mathrm{\Σ}}}{T}_{\mathrm{adp}}\left(k\right)$

In the formula, c is the highway structure critical parameter, c ∈ [2,3], [K ₁, K ₂] for having comprised the frequency band of track full detail.

5. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 4, its characteristic is: described step 3 further comprises:

Effectively vehicle power spectrum normalization;

The normalized power accumulation.

6. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 5 is characterized in that:

Following condition is satisfied in the normalization of described effective vehicle power spectrum:

$\left\{\begin{array}{c}{P}_{\mathrm{vnorm}}\left(k\right)=\frac{(P_R\left(k\right)-T_{\mathrm{adp}}\left(k\right))\·g\left(k\right)}{\max \left(T_{\mathrm{adp}}\right)}\\ \underset{k=K_1}{\overset{K_2}{\mathrm{\Σ}}}{P}_{\mathrm{vnorm}}\left(k\right)\≠0\end{array}\right.,$

In the formula, P _VnormBe effective vehicle normalized power spectrum, P _Vnorm(k) be the value of effective vehicle normalized power spectrum, P at frequency k place _R(k) be the power spectrum at frequency k place, max (T _Adp) expression T _Adp(k) maximal value in, Z _AdpBe adaptive threshold, k ∈ [K ₁, K ₂], [K ₁, K ₂] be the frequency band that has comprised the track full detail,

G (k) is effective vehicle discriminant function:

7. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 6 is characterized in that:

Described normalized power accumulation can be expressed from the next:

$\left\{\begin{array}{c}{P}_{\ln \mathrm{orm}}^n\left(k\right)=\frac{(n-1)P_{\ln \mathrm{orm}}^{n-1}\left(k\right)+P_{\mathrm{vnorm}}\left(k\right)}{n}\\ P_{\ln \mathrm{orm}}^0\left(k\right)=T_{\mathrm{adp}}\left(k\right)\end{array}\right.n=1,...,N,$

In the formula, P _LnormBe the normalized power cumulative function, n is the sample size that is used for current lane identification, k ∈ [K ₁, K ₂], [K ₁, K ₂] for having comprised the frequency band of track full detail, P _VnormBe effective vehicle normalized power spectrum, T _Adp(k) be frequency k place adaptive threshold.

8. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 7 is characterized in that:

Described level and smooth normalization cumulative power spectrum realizes that by the smoothing windows function smoothing windows function uses rectangular window:

$w_k\left(n\right)=u(n-k+\frac{W-1}{2})-u(n-k-\frac{W-1}{2}),$

In the formula, u (n) is a step function, and window width W is an odd number, k ∈ [K ₁, K ₂], [K ₁, K ₂] for having comprised the frequency band of track full detail;

Normalization cumulative power spectrum after level and smooth can be expressed from the next:

$\stackrel{\‾}{P_{\ln \mathrm{orm}}\left(k\right)}=\frac{\underset{i=K_1}{\overset{K_2}{\mathrm{\Σ}}}(P_{\ln \mathrm{orm}}\left(i\right)\·w_k\left(i\right))}{W}.$

9. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 1 is characterized in that:

The current road driveway of described identification border, need border and forward and reverse highway border between the identification road driveway, if highway track forward or backwards comprises n bar track, number of boundary is n-1 between the road driveway that needs so to determine, the highway number of boundary is 2 forward or backwards.

10. the automatic recognition method of road driveway based on millimeter wave traffic radar as claimed in claim 1 is characterized in that:

The described self study process of utilizing realizes road driveway identification, is meant that at each frequency place, the self study process of lane identification can be expressed as:

$L^t\left(k\right)=(1-r)L^{t-1}\left(k\right)+r{L}_0^t\left(k\right)$

L is a t road driveway recognizing site constantly in the formula, L ₀Be current lane identification position, r is a learning rate, k ∈ [K ₁, K ₂], [K ₁, K ₂] for having comprised the frequency band of track full detail.

Images