Computer implemented method and related system for extraction of characteristic parameters of traffic
The method employs gradient-square tensor analysis and filtering techniques on spatio-temporal data from optical fiber cables to accurately determine vehicle velocity, addressing limitations in existing DAS systems and improving accuracy in dense traffic scenarios.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- TA-MONITORING
- Filing Date
- 2025-01-23
- Publication Date
- 2026-07-30
AI Technical Summary
Existing distributed acoustic sensing (DAS) systems face challenges in accurately determining vehicle velocity, especially in high-density traffic conditions, due to insufficient differentiation between vehicle traces and limitations in linearity assumptions, leading to inaccurate velocity calculations and peak detection in dense traffic scenarios.
A computer-implemented method using a gradient-square tensor analysis on spatio-temporal data images derived from optical fiber cable vibrations, combined with filtering techniques and noise reduction, to accurately determine vehicle velocity over a wide range of speeds.
The method enables precise determination of vehicle velocity across a wide range of speeds, effectively filtering noise and non-relevant data, enhancing accuracy in high-density traffic conditions.
Smart Images
Figure EP2025051688_30072026_PF_FP_ABST
Abstract
Description
[0001] Computer implemented method and related system for extraction of characteristic parameters of traffic
[0002] Technical field
[0003] The present invention relates to a computer implemented method and related system for extraction of characteristic parameters of traffic, preferably the velocity of the traffic.
[0004] Background art
[0005] Distributed acoustic sensing, DAS, systems are sensing systems using optical fiber cables as sensing element and transmission signal medium. Such systems mainly include three parts, i.e. a sensing optical fiber cable, a pulse generating device and a receiver. The pulse generating device is configured for transmitting coherent light pulses through, i.e. an optical signal, through the sensing optical fiber cable. While propagating through the sensing optical fiber cable the light pulses are reflected back, i.e. backscattered, at positions where the sensing optical fiber cable is strained, for example due to acoustic vibrations in the sensing optical fiber cable. The reflected light pulses return along the sensing optical fiber cable back to the origin, where they are received by the receiver.
[0006] This technology may be applicable for measuring signal change caused by past vehicle vibration, so that traffic information such as traffic flow, average speed, vehicle number and the like can be detected.
[0007] Each point of the fiber cable corresponds to a sensor node, providing neartime continuous high-speed environmental sampling, and each distributed sensor node cooperates to sense the vibration signal on the whole line. The position of road surface vibration caused when a vehicle runs on a highway varies with time, and thus the acquired vehicle vibration signal appears as a spatiotemporal two-dimensional image. In the context of highway monitoring, the vehicle speed is almost constant over a relatively short period of time and under free traffic conditions, the vehicle travel trajectory can be approximated as a straight line on a spatio-temporal two-dimensional map of the DAS signal, the slope of which can be understood as the vehicle speed, the direction of the slope representing the direction of travel of the vehicle. Moreover, the smaller and lighter the vehicle, the farther from the optical fiber cable, the smaller the vibration intensity generated,and the more blurred the space-time two-dimensional vibration profile formed by the DAS signal, and even the more obstructed by other noise.
[0008] The high-speed traffic monitoring system provides information about traffic conditions, such as the number and the speed of vehicles, can potentially reflect traffic situations such as traffic jams, accidents and the like, and has importance and profound significance in the aspects of improving traffic efficiency, enhancing safety, optimizing resource utilization, promoting traffic technology innovation and the like.
[0009] A first known approach wherein the velocity of vehicles is estimated in DAS data uses a Hough Transform algorithm which tries to draw lines on the DAS plot and extract velocity measures from the slope of these lines. This approach has two significant shortcomings. Firstly, when the density of traffic is high, the traces of vehicles close to each other cannot be differentiated sufficiently to provide satisfying accuracy. Secondly the traces of vehicles do not necessarily follow a determined shape like a line or a parameterizable curve such as a circle or ellipse, in a time-position visualization. Vehicles are accelerating and decelerating and therefore create a spontaneous trace in the DAS energy image. This circumstance limits the application of existing approaches to the detection of linear traces only.
[0010] A further known approach uses the characteristics of traffic vibration data and applies a wavelet threshold algorithm for data filtering and an improved dualthreshold algorithm on the short-term energy and zero crossing rate of the strain. The approach overcomes the linearity or constant velocity constraint of the Hough Transform. However, the velocity calculation only provides acceptable results for velocities less than 40 km / h. The spread in the strain signal for higher velocities is too high and the generalization to dense traffic is not feasible. Furthermore, this approach does not provide sufficient accuracy for counting. The detection of peaks (cars) in case several cars are following each other in short distance cannot be differentiated in an acceptable manner.
[0011] Disclosure of the invention
[0012] An object of the present invention is to provide a computer implemented method and related system for extracting characteristic parameters of traffic, preferably the velocity of the traffic, wherein the aforementioned shortcoming or drawbacks of the known solutions are alleviated or overcome.According to the present invention, this object is achieved by the method as described in the first independent claim and the system as described in the second independent claim.
[0013] Therefore, the present invention provides a computer implemented method for extraction of characteristic parameters of traffic. The traffic comprises at least one moving vehicle on a road. The method comprises the step of transmitting, by a pulse generating device, an optical signal, i.e. plurality of light pulses, into a sensing optical fiber cable positioned along, preferably parallel to, said road. Preferably, the pulse generating device transmits the optical signal into the sensing optical fiber cable at a first end of the sensing optical fiber cable. Preferably, the pulse generating device comprises a laser in combination with a pulse generator, optionally followed by an optical amplifier. The method comprises the step of receiving, by a receiver, from said sensing optical fiber cable a reflected optical signal resulting from acoustic vibrations in said sensing optical fiber cable caused by said at least one moving vehicle. Preferably, the receiver receives the reflected optical signal at the first end of said sensing optical fiber cable. The reflected optical signal may be received for each individual light pulse of the optical signal, but the reflected optical signal may also be received for a plurality of light pulses of the optical signal over a predetermined period of time and combined with each other. The latter may for example be done with a 1 D Fourier Transform Gaussian filter by gathering the reflections from different light pulses, performing a 1D Fourier Transform, for instance a 1 D Fast Fourier Transform, on the gathered signal in the time domain, and add the signal intensity over all frequencies, or a sub-selection of frequencies of interest, for each of the positions along the sensing optical fiber cable. It should however be clear that any other suitable method known to the skilled person for determining the signal intensity at each position and time step can be used, such as for example standard deviation or variance of the signal, power spectral density, etc. The method comprises the step of deriving, by at least one processing unit, a spatio-temporal data image d, or spatio-temporal two-dimensional map, from said reflected optical signal. The at least one processing unit may be part of the receiver. The at least one processing unit may also be part of a computing device separate from the receiver, such as for example a local computer at the location of the receiver or a remote server at a remote location from the receiver. The at least one processing unit may be at least two processingunits divided over the receiver and a computing device separate from the receiver, wherein for example initial or receiver specific processing is performed by the processing unit of the receiver and further processing or more generic processing is performed by the processing unit of the computing device. The method comprises the step of determining, by said at least one processing unit, for at least one pixel, preferably for at least one subset of adjacent pixels, preferably for each pixel, in said generated spatio-temporal data image d a gradient-square tensor, or structure tensor, G. Preferably, the gradient-square tensor is determined as G = Herein, the
[0014]
[0015] parameter t refers to time and the parameter x to the position along the sensing optical fiber cable. Herein,
[0016]
[0017] may simply be calculated as the difference between the values of said generated spatio-temporal data image d at said at least one pixel and an adjacent pixel along the direction t divided by the time difference between said adjacent pixels, and
[0018]
[0019] as the difference between the values of said generated spatio-temporal data image d at said at least one pixel and an adjacent pixel along the direction x divided by the position difference between said adjacent pixels along the sensing optical fiber cable. It should however be clear that any other suitable numerical method known to the skilled person for calculating said derivatives can be used. The method comprises the step of determining, by said at least one processing unit, a slope p for said at least one pixel of said spatio-temporal data image d from the gradient-square tensor G. Preferably, the slope p is determined from the largest eigenvalue of the gradient-square tensor G. Preferably, the slope is determined as whereinmaxis the largest eigenvalue of the
[0020]
[0021] gradient-square tensor G. The method comprises the step of converting, by said at least one processing unit, said slope p into a velocity of the corresponding vehicle on the road by means of a linear transformation of said slope. The linear transformation may for example be performed by multiplying said slope p with the position resolution dx of the spatio-temporal data image d divided by the time resolution dt of the spatio-temporal data image d.
[0022] The method according to the present invention offers the advantage that the slope of a trace in the spatio-temporal data image corresponding to a vehicle is determined accurately along the length of said trace, and this over a wide rangeof slopes. Hence, this allows the velocity of the corresponding vehicle to be determined accurately over a wide range of velocities, and allows the changes in the velocity of the corresponding vehicle over time to be determined accurately.
[0023] In an embodiment of the method according to the present invention the step of determining the gradient-square tensor G further comprises applying a predetermined smoothing filter on the components of the gradient-square tensor G, such as for example the components gtt> 9xxandgtxof the the gradient-square [9
[0024] ntt 9
[0025] ntx l
[0026] . The predetermined smoothing filter may for example be a Gaussian filter, but any other suitable smoothing filter known to the skilled person can be used. It should be noted that for being able to apply such a smoothing filter, the components of the gradient-square tensor G, and thus the gradient-square tensor G, should have been determined for at least a subset of adjacent pixels of the spatio-temporal data image d, preferably in both directions of the spatiotemporal data image d. Preferably, the components of the gradient-square tensor G, and thus the gradient-square tensor G, are thereby determined for each pixel of the spatio-temporal data image d.
[0027] The smoothing is beneficial for smoothing out local changes in these gradients that are not representative for the local slope of a trace in the spatiotemporal data image corresponding to a vehicle. In this manner the velocity of said corresponding vehicle is determined more accurately.
[0028] In an embodiment of the method according to the present invention the method further comprises the step of transforming, by said said at least one processing unit, said spatio-temporal data image d into a frequency-wavenumber, F-K, domain representation. This may be done by performing a 2D Fourier Transform, for instance a 2D Fast Fourier Transform, on the spatio-temporal data image d. The method further comprises the step of applying, by said at least one processing unit, a predetermined F-K filter on the F-K domain representation configured for filtering out frequencies and / or wavenumbers outside a predetermined velocity range. This can be done easily in the F-K domain representation since the velocity is a ratio of the frequency and the wavenumber. The predetermined F-K filter may for example be implemented by setting all pixels in the F-K domain representation for which the corresponding velocity, i.e. the ratio of the frequency and the wavenumber, is outside of the predetermined velocityrange to zero. A sinusoidal roll off may however be implemented in the predetermined F-K filter. The method further comprises the step of inverse transforming, by said at least one processing unit, said F-K domain representation to the spatio-temporal domain. Thereby, the spatio-temporal data image d is obtained in velocity filtered form.
[0029] This embodiment is beneficial for filtering data relating to static non-moving objects out of the spatio-temporal data image which are not of interest. This embodiment is also beneficial for filtering data relating to high speed objects out of the spatio-temporal data image which are not of interest. Such data relating to high speed objects may for example be vibrations caused seismic waves. This embodiment is also beneficial for filtering data relating to moving vehicles out of the spatio-temporal data image that are moving in a direction opposite to that of the vehicles of interest.
[0030] In an embodiment of the method according to the present invention the predetermined velocity range ranges from 10 km / h to 220 km / h. Preferably, the predetermined velocity range ranges from 12 km / h to 210 km / h. More preferably, the predetermined velocity range ranges from 14 km / h to 200 km / h. Even more preferably, the predetermined velocity range ranges from 16 km / h to 190 km / h. Yet even more preferably, the predetermined velocity range ranges from 18 km / h to 180 km / h.
[0031] In an embodiment of the method according to the present invention the step of receiving the reflected optical signal further comprises thresholding, by said at least one processing unit, the reflected optical signal to filter out sudden spikes in the reflected optical signal.
[0032] This embodiment offers the advantage that any sudden spikes in the reflected optical signal that don't seem to fit based on the total strain observed are filtered out to improve the quality of the received reflected optical signal resulting from acoustic vibrations in the sensing optical fiber cable.
[0033] In an embodiment of the method according to the present invention the method further comprises the step of determining a noise level. The step of determining the noise level comprises transmitting, by the pulse generating device, an optical signal into a reference optical fiber cable. Preferably, the reference optical fiber cable is insulated from acoustic vibrations. The reference optical fiber cable may be a reference section of the sensing optical fiber cable. The referenceoptical fiber cable may also be an optical fiber cable separate from the sensing optical fiber cable. The step of determining the noise level comprises receiving, by the receiver, from said reference optical fiber cable a reflected reference optical signal. The step of determining the noise level comprises determining, by said at least one processing unit, the noise level from the reflected reference optical signal. The step of receiving the reflected optical signal further comprises subtracting, by said at least one processing unit, the determined noise level from the reflected optical signal.
[0034] The noise level may for example be determined by measuring the reflected reference optical signal at different time steps, calculating the mean over segments of the reflected reference optical signal at each time step, subtracting from said means the mean of a corresponding segment of the reflected reference optical signal at a preceding time step, and calculating the median over the subtracted means. It should however be noted that any other suitable method for determining a noise level from a signal known the skilled person can be used.
[0035] This embodiment is beneficial for removing noise from the reflected optical signal and thus from the spatio-temporal data image, which noise could distort the determination of the slope of a trace in the spatio-temporal data image corresponding to a vehicle. In this manner the velocity of said corresponding vehicle is determined more accurately.
[0036] In an embodiment of the method according to the present invention the step of receiving the reflected optical signal further comprises converting, by said at least one processing unit, the reflected optical signal to nanostrain, nstrain, by means of a look-up table specific to the pulse generating device, the sensing optical fiber cable and the receiver.
[0037] This embodiment is beneficial for standardizing the measurement values of the received reflected optical signal. This allows the further steps of the method to be implemented independent of the used pulse generating device, sensing optical fiber cable and receiver.
[0038] In an embodiment of the method according to the present invention the step of receiving the reflected optical signal further comprises applying, by said at least one processing unit, a predetermined bandpass filter to the reflected optical signal for filtering out frequencies outside of a predetermined frequency range.This embodiment is beneficial for filtering frequencies out of the reflected optical signal that are irrelevant for traffic monitoring. Removing irrelevant frequencies improves the signal-to-noise ratio of the received reflected optical signal resulting from acoustic vibrations. The filter that is applied is a bandpass filter which defines a low cut-off frequency where everything below that frequency is removed and a high cut-off frequency where everything above that frequency is removed.
[0039] In an embodiment of the method according to the present invention the predetermined frequency range ranges from 0,1 Hz to 4,0 Hz. Preferably, the predetermined frequency range ranges from 0,2 Hz to 3,5 Hz. More preferably, the predetermined frequency range ranges from 0,3 Hz to 3,0 Hz. Even more preferably, the predetermined frequency range ranges from 0,4 Hz to 2,5 Hz. Yet even more preferably, the predetermined frequency range ranges from 0,5 Hz to 2,0 Hz.
[0040] It is found that acoustic vibrations in this predetermined frequency range are caused by the noise of the tires of the vehicles moving on the road, which are the acoustic vibrations that are of interest for traffic monitoring.
[0041] Typically, a bandpass filter between 0,5 Hz and 2,0 Hz is applied as this bandwidth from the reflected optical signal resulting from acoustic vibrations appears to be most relevant for purposes for extraction of characteristic parameters of traffic.
[0042] In an embodiment of the method according to the present invention the step of determining the slope p further comprises applying an error correction on the slope p based on a slope error retrieved from synthetic examples with a known slope.
[0043] By applying the method on synthetic examples, i.e. computer generated spatio-temporal data images comprising traces with known slopes, the inventors have found that the method over-predicts the slope for slopes below TT / 4 radians and under-predicts the slope for slopes above TT / 4 radians. The curve of this error may be fitted by a high degree polynomial, which can beneficially be used for correcting the slope of a trace in the spatio-temporal data image corresponding to a vehicle. In this manner the velocity of said corresponding vehicle is determined more accurately.The present invention also provides a system for extraction of characteristic parameters of traffic data patterns from traffic. The traffic comprises at least one moving vehicle on a road. The system comprises a sensing optical fiber cable positioned along, preferably parallel to, said road. The system comprises a pulse generating device configured for transmitting an optical signal into said sensing optical fiber cable. Preferably, the pulse generating device is positioned at a first end of the sensing optical fiber cable. The system comprises a receiver configured receiving a reflected optical signal resulting from acoustic vibrations in said sensing optical fiber cable caused by said at least one moving vehicle. Preferably, the receiver is positioned at the first end of the sensing optical fiber cable. The system comprises at least one processing unit. The at least one processing unit is configured for deriving a spatio-temporal data image from said reflected optical signal. The at least one processing unit is configured for determining for at least one pixel in said spatio-temporal data image a gradient-square tensor G. The at least one processing unit is configured for determining a slope p for said at least one pixel of said spatio-temporal data image from the gradient-square tensor G. The at least one processing unit is configured for converting said slope p into a velocity of the corresponding vehicle on the road by means of a linear transformation of said slope p.
[0044] The present invention also provides a computing device for a system for extraction of characteristic parameters of traffic data patterns from traffic. The traffic comprises at least one moving vehicle on a road. The system comprises a sensing optical fiber cable positioned along, preferably parallel to, said road. The system comprises a pulse generating device configured for transmitting an optical signal into said sensing optical fiber cable. Preferably, the pulse generating device is positioned at a first end of the sensing optical fiber cable. The system comprises a receiver configured receiving a reflected optical signal resulting from acoustic vibrations in said sensing optical fiber cable caused by said at least one moving vehicle. Preferably, the receiver is positioned at the first end of the sensing optical fiber cable. The computing device comprises at least one processing unit. The at least one processing unit is configured for deriving a spatio-temporal data image d from said reflected optical signal. The at least one processing unit is configured for determining for at least one pixel in said spatio-temporal data image d a gradientsquare tensor G. The at least one processing unit is configured for determining aslope p for said at least one pixel of said spatio-temporal data image d from the gradient-square tensor G. The at least one processing unit is configured for converting said slope p into a velocity of the corresponding vehicle on the road by means of a linear transformation of said slope p.
[0045] Brief description of the drawings
[0046] The invention will be further elucidated by means of the following description and the appended figures.
[0047] Figure 1 shows a schematical representation of a distributed acoustic sensing system according to an embodiment of the present invention.
[0048] Figures 2A and 2B show a received reflected optical signal respectively before and after application of a bandpass filter.
[0049] Figures 3A and 3B show a visualization of traffic data in the F-K domain before and after application of a F-K filter configured for filtering out frequencies and / or wavenumbers outside a predetermined velocity range.
[0050] Figures 4A, 4B and 4C illustrates the directionality filtering using a F-K filter. Figure 5 illustrates the noise level estimation from a reflected reference optical signal from a reference optical fiber cable.
[0051] Figure 6 shows a plot of the slope measurement error.
[0052] Modes for carrying out the invention
[0053] The present invention will be described with respect to particular embodiments and with reference to certain drawings but the invention is not limited thereto but only by the claims. The drawings described are only schematic and are non-limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn on scale for illustrative purposes. The dimensions and the relative dimensions do not necessarily correspond to actual reductions to practice of the invention.
[0054] Furthermore, the terms first, second, third and the like in the description and in the claims, are used for distinguishing between similar elements and not necessarily for describing a sequential or chronological order. The terms are interchangeable under appropriate circumstances and the embodiments of the invention can operate in other sequences than described or illustrated herein.Moreover, the terms top, bottom, over, under and the like in the description and the claims are used for descriptive purposes and not necessarily for describing relative positions. The terms so used are interchangeable under appropriate circumstances and the embodiments of the invention described herein can operate in other orientations than described or illustrated herein.
[0055] The term “comprising”, used in the claims, should not be interpreted as being restricted to the means listed thereafter; it does not exclude other elements or steps. It needs to be interpreted as specifying the presence of the stated features, integers, steps or components as referred to, but does not preclude the presence or addition of one or more other features, integers, steps or components, or groups thereof. Thus, the scope of the expression “a device comprising means A and B” should not be limited to devices consisting only of components A and B. It means that with respect to the present invention, the only relevant components of the device are A and B.
[0056] In the following paragraphs, referring to the drawing in Figure 1, an implementation of the system 100 for extraction of characteristic parameters of traffic according to an embodiment of the present invention is described.
[0057] The system 100 comprises a sensing optical fiber cable 130 that stretches along a road to monitor. In order to record the traffic data, the sensing optical fiber cable 130 is arranged along the road. This can be done by placing it freely on the roadside, mounting on or integrating it into or attaching it onto a vehicle restraint system along the road, such as a steel safety barrier or concrete protective wall, or in a similar way.
[0058] In order to increase the sensitivity of the sensing optical fiber cable 130 it is proposed to sense the structure-born sound of tires. This can be achieved by attaching the cable directly or indirectly to the road surface. To increase the sensitivity for structure borne sound the sensing optical fiber cable 130 can be covered with a durable material like asphalt. Furthermore, the sensing optical fiber cable 130 can be embedded directly inside the road surface.
[0059] This approach can be strengthened by embedding the sensing optical fiber cable 130 close to the typical tire track location. Because a structural material is a better transmitter of acoustic waves respectively vibration, an embedded sensing optical fiber cable 130 is much more sensitive to the noise of a rolling tire and therefore able to detect even the slowest tires rolling on the road surface.Furthermore, it is proposed to embed several sensing optical fiber cables 130 along the road, preferably one sensing optical fiber cable 130 on each edge of the road or at each traffic lane, in order to be able to detect and separate vehicles 200 on several traffic lanes on a road or highway. Furthermore, by embedding the sensing optical fiber cable 130 in the ground respectively in the road surface respectively in a road safety barrier the reduction or elimination of the airborne sound will also reduce or eliminate blur from the data output.
[0060] The system 100 further comprises a pulse generating device 110, which comprises a laser 111 in combination with a pulse generator 112. In combination, these devices are configured to generate an optical signal 310 comprising a series of light pulses for transmitting into the sensing optical fiber 130 being positioned along the road whereon traffic is monitored.
[0061] The pulse generating device 110 may further comprise an optical amplifier 113 that is configured to amplify the generated optical signal 310 to the appropriate amplitude for optimum adapting the optical signal 310 to the sensing optical fiber cable 130.
[0062] The system 100 also comprises a receiver 120 that is configured to receive a reflected optical signal 320 at the first end of the sensing optical fiber cable 130. This reflected optical signal 320 is the result of backscattering of the light pulses of the optical signal 310 in the sensing optical fiber cable 130 at positions where the sensing optical fiber cable 130 is strained due to acoustic vibrations in the sensing optical fiber cable 130, such as acoustic vibrations caused by said at least one moving vehicle 200 on the road.
[0063] The receiver 120 may comprise a receiving component 121 that is configured specifically for receiving the reflected optical signal 320, and a processing unit 122 that is configured for performing further data processing on the reflected optical signal 320 received by the receiving component 121, such as the processing according to embodiments of the method according to the present invention. In alternative embodiments the processing unit 122 may also be part of a computing device separate from the receiver 120, such as for example a local computer at the location of the receiver 120 or a remote server at a remote location from the receiver 120. In alternative embodiments the processing unit 122 may be provided in the receiver 120 and a further processing unit in such a computing device separate from the receiver 120. Thereby, for example, the processing unit122 of the receiver may be used for initial or receiver 120 specific processing, and the processing unit of the computing device may be used for further or more generic processing.
[0064] In the succeeding paragraph the method performed by the system 100 for extraction of characteristic parameters of traffic according to an embodiment of the present invention is described.
[0065] In order to explain an embodiment of the present invention it is assumed that the traffic on a road comprises at least one moving vehicle 200 on a road and it is further assumed that a pulse generating device 110, generates an optical signal 310, that comprises a plurality of light pulses which are transmitted by means of the pulse generating device 110 into the coupled sensing optical fiber cable 130 positioned along, preferably parallel to, said road.
[0066] The receiver 120 coupled to a first end of said sensing optical fiber cable 130 receives a reflected optical signal 320 resulting from acoustic vibrations in said optical fiber cable 130 caused by said at least one moving vehicle 200. The reflected optical signal 320 received for each individual light pulse of the optical signal 310 transmitted may be received and processed separately, but the reflected optical signal 320 may also be received for a plurality of light pulses of the optical signal 310 over a predetermined period of time and combined with each other. The latter may for example be done with a 1 D Fourier Transform Gaussian filter by the processing unit 122 by gathering the respective reflections from different transmitted light pulses, performing a 1 D Fourier Transform, for instance a 1 D Fast Fourier Transform, on the gathered signal in the time domain, and sum the obtained signal intensity over all frequencies, or a sub-selection of frequencies of interest, for each of the positions along the sensing optical fiber cable 130. Such a sub-selection of frequencies of interest may for example be any one of the frequency ranges specified for the predetermined bandpass filter, which will be discussed further below. It should however be clear that any other suitable method known to the skilled person for determining the signal intensity at each position and time step can be used, such as for example standard deviation or variance of the signal, power spectral density, etc. The method further comprises the step of deriving, by said processing unit 122, a spatio-temporal data image d, or spatiotemporal two-dimensional map, from said reflected optical signal 320 received. An example of such a spatio-temporal data image can be seen in Figure 3A.The method further comprises the step of determining, by said processing unit 122, for at least one pixel, preferably for each pixel in at least one subset of adjacent pixels, preferably for each pixel, in said generated spatio-temporal data
[0067] , wherein
[0068]
[0069] 9tt =
[0070]
[0071] and gtx= ^ - ^- Herein, the parameter t refers to time and the parameter x to the position along the sensing optical fiber cable 130.
[0072]
[0073] Herein, may simply be calculated as the difference between the values of said generated spatio-temporal data image d at said at least one pixel and an adjacent pixel along the direction t divided by the time difference between said adjacent
[0074]
[0075] pixels, and
[0076]
[0077] as the difference between the values of said generated spatiotemporal data image d at said at least one pixel and an adjacent pixel along the direction x divided by the position difference between said adjacent pixels along the sensing optical fiber cable. It should however be clear that any other suitable
[0078]
[0079] numerical method known to the skilled person for calculating said derivatives
[0080]
[0081]
[0082] can be used.
[0083] The method further comprises the step of determining, by said processing unit 122, a slope p =max~9ttfor said at least one pixel of said spatio-temporal data 9tx
[0084] image d, whereinmaxis the largest eigenvalue of the gradient-square tensor G. The method comprises the subsequent step of converting, by said receiver 120, said slope p into a velocity v of the corresponding vehicle 200 on the road by means of a linear transformation of said slope. The linear transformation may for example be performed by multiplying said slope p with the position resolution dx of the spatio-temporal data image d divided by the time resolution dt of the spatiotemporal data image d, i.e. v = p ^.
[0085] The method according to the present invention offers the advantage that the slope of a trace in the spatio-temporal data image d corresponding to a vehicle 200 on the road is determined accurately along the length of said trace, and this over a wide range of slopes. Hence, this allows the velocity of the corresponding vehicle 200 to be determined accurately over a wide range of velocities, and allowsthe changes in the velocity of the corresponding vehicle 200 over time to be determined accurately.
[0086] In an advantageous embodiment of the method according to the present invention the step of determining the gradient-square tensor G further comprises applying a predetermined smoothing filter on the components gt,gxxand gtxof the gradient-square tensor G. The predetermined smoothing filter may for example be a Gaussian filter, but any other suitable smoothing filter known to the skilled person can be used. It should be noted that for being able to apply such a smoothing filter, the components gt,gxxand gtxof the gradient-square tensor G should have been determined for at least a subset of adjacent pixels of the spatiotemporal data image d, preferably in both directions of the spatio-temporal data image d. Preferably, the components gtt> gxxand gtxof the gradient-square tensor G are thereby determined for each pixel of the spatio-temporal data image d.
[0087] The smoothing is beneficial for smoothing out local changes in these gradients that are not representative for the local slope of a trace in the spatiotemporal data image d corresponding to a vehicle 200. In this manner the velocity of said corresponding vehicle 200 is determined more accurately.
[0088] In a further embodiment of the method according to the present invention the method further comprises the step of transforming, by said processing unit 122, said spatio-temporal data image d into a frequency-wavenumber, F-K, domain representation. This may be done by performing a 2D Fourier Transform, for instance, a 2D Fast Fourier Transform, applied on the spatio-temporal data image d. The method further comprises the step of applying, by said processing unit 122, a predetermined F-K filter on the F-K domain representation configured for filtering out frequencies and / or wavenumbers outside a predetermined velocity range. This can be done easily in the F-K domain representation since the velocity is a ratio of the frequency and the wavenumber. The predetermined F-K filter may for example be implemented by setting all pixels F-K domain representation for which the corresponding velocity, i.e. the ratio of the frequency and the wavenumber, is outside of the predetermined velocity range to zero. A sinusoidal roll off may however be implemented in the predetermined F-K filter. The method further comprises the step of inverse transforming, by said processing unit 122, said F-K domain representation to the spatio-temporal domain. Thereby, the spatiotemporal data image d is obtained in velocity filtered form.This filtering is illustrated by Figures 2A and 2B, which respectively show the visualization of 2 minutes traffic data in the F-K domain without filtering and with filtering.
[0089] This embodiment is beneficial for filtering data relating to static non-moving objects out of the spatio-temporal data image d which are not of interest. Data relating to such objects can be seen along the vertical axis in Figure 2A. This embodiment is also beneficial for filtering data relating to high-speed objects out of the spatio-temporal data image d which are not of interest. Such data relating to high-speed objects may for example be vibrations caused seismic waves. Data relating to such objects can be seen along the horizontal axis in Figure 2A. This embodiment is also beneficial for filtering data relating to moving vehicles 200 out of the spatio-temporal data image d that are moving in a direction opposite to that of the vehicles 200 of interest. Data relating to vehicles 200 moving in one direction can be seen along a diagonal from top left to bottom right in Figure 2A. Data relating to vehicles 200 moving in the opposite direction can be seen along a diagonal from bottom left to top right in Figure 2A.
[0090] The final result of such filtering, i.e. after inverse transforming the F-K domain representation, is illustrated by Figures 3A, 3B and 3C. Figure 3A shows the spatio-temporal image before filtering. This spatio-temporal image shows forwards tilted traces associated with vehicles 200 moving in a forwards direction, and backwards tilted traces associated with vehicles 200 moving in a reverse direction. Figure 3B shows the spatio-temporal image after filtering, in which only the forwards tilted traces associated with the vehicles 200 moving in the forwards direction are retained. The method according to the present invention for determining the velocity of the associated vehicles 200 is performed on the spatiotemporal image of Figure 3B. Figure 3C shows the spatio-temporal image of the data that is filtered out, i.e. the backwards tilted traces associated with the vehicles 200 moving in the reverse direction. It should be noted that, if desired, the method according to the present invention for determining the velocity of the associated vehicles 200 may also be performed on the spatio-temporal image of Figure 3C.
[0091] In an embodiment of the method according to the present invention the predetermined velocity range ranges from 10 km / h to 220 km / h. Preferably, the predetermined velocity range ranges from 12 km / h to 210 km / h. More preferably, the predetermined velocity range ranges from 14 km / h to 200 km / h. Even morepreferably, the predetermined velocity range ranges from 16 km / h to 190 km / h. Yet even more preferably, the predetermined velocity range ranges from 18 km / h to 180 km / h.
[0092] In an embodiment of the method according to the present invention the step of receiving the reflected optical signal 320 further comprises thresholding, by the processing unit 122, the reflected optical signal to filter out sudden spikes in the reflected optical signal.
[0093] This embodiment offers the advantage that any sudden spikes in the reflected optical signal 320 that don't seem to fit based on the total strain observed are filtered out to improve the quality of the received reflected optical signal 320 resulting from acoustic vibrations in the sensing optical fiber cable 130.
[0094] In an embodiment of the method according to the present invention the method further comprises the step of determining a noise level. The step of determining the noise level comprises transmitting, by the pulse generating device 110, an optical signal into a reference optical fiber cable. Preferably, the reference optical fiber cable is insulated from acoustic vibrations. The reference optical fiber cable may be a reference section of the sensing optical fiber cable 130. The reference optical fiber cable may also be an optical fiber cable separate from the sensing optical fiber cable 130. The step of determining the noise level comprises receiving, by the receiver 120 at a first end of said reference optical fiber cable, a reflected reference optical signal. The step of determining the noise level comprises determining, by the processing unit 122, the noise level from the reflected reference optical signal. The step of receiving the reflected optical signal 320 further comprises subtracting the determined noise level from the reflected optical signal 320.
[0095] The noise level may for example be determined, as illustrated in Figure 5, by measuring the reflected reference optical signal at different time steps, calculating the mean over segments of the reflected reference optical signal at each time step, subtracting from said means the mean of a corresponding segment of the reflected reference optical signal at a preceding time step, and calculating the median over the subtracted means. It should however be noted that any other suitable method for determining a noise level from a signal known the skilled person can be used.This embodiment is beneficial for removing noise from the reflected optical signal 320 and thus from the spatio-temporal data image, which noise could distort the determination of the slope of a trace in the spatio-temporal data image corresponding to a vehicle 200. In this manner the velocity of said corresponding vehicle 200 is determined more accurately.
[0096] In an embodiment of the method according to the present invention the step of receiving the reflected optical signal 320 further comprises converting, by the processing unit 122, the reflected optical signal 320 to nanostrain, nstrain, by means of a look-up table specific to the pulse generating device 110, the sensing optical fiber cable 130 and the receiver 120.
[0097] This embodiment is beneficial for standardizing the measurement values of the received reflected optical signal 320. This allows the further steps of the method to be implemented independent of the used pulse generating device 110, sensing optical fiber cable 130 and receiver 120.
[0098] In an embodiment of the method according to the present invention the step of receiving the reflected optical signal 320 further comprises applying, by said processing unit 122, a predetermined bandpass filter to the reflected optical signal 320 for filtering out frequencies outside of a predetermined frequency range.
[0099] This embodiment is beneficial for filtering frequencies out of the reflected optical signal 320 that are irrelevant for traffic monitoring. Removing irrelevant frequencies improves the signal-to-noise ratio of the received reflected optical signal 320 resulting from acoustic vibrations. The filter that is applied is a bandpass filter which defines a low cut-off frequency where everything below that frequency is removed and a high cut-off frequency where everything above that frequency is removed.
[0100] In an embodiment of the method according to the present invention the predetermined frequency range ranges from 0,1 Hz to 4,0 Hz. Preferably, the predetermined frequency range ranges from 0,2 Hz to 3,5 Hz. More preferably, the predetermined frequency range ranges from 0,3 Hz to 3,0 Hz. Even more preferably, the predetermined frequency range ranges from 0,4 Hz to 2,5 Hz. Yet even more preferably, the predetermined frequency range ranges from 0,5 Hz to 2,0 Hz.It is found that acoustic vibrations in this predetermined frequency range are caused by the noise of the tires of the vehicles 200 moving on the road, which are the acoustic vibrations that are of interest for traffic monitoring.
[0101] Typically, a bandpass filter between 0,5 Hz and 2 Hz is applied as this bandwidth from the reflected optical signal 320 resulting from acoustic vibrations appears to be most relevant for purposes for extraction of characteristic parameters of traffic as is shown in respective figures 4A and 4B showing a received reflected optical signal 320 without additional bandpass filtering and a received reflected optical signal 320 with application of a bandpass filter.
[0102] In an additional embodiment of the method according to the present invention the step of determining the slope p further comprises applying an error correction on the slope p based on a slope error retrieved from synthetic examples with a known slope.
[0103] By applying the method on synthetic examples, i.e. computer generated spatio-temporal data images comprising traces with known slopes, the inventors have found, as illustrated in Figure 6, that the method over-predicts the slope for slopes below TT / 4 radians and under-predicts the slope for slopes above TT / 4 radians. The curve of this error may be fitted by a high degree polynomial, which can beneficially be used for correcting the slope of a trace in the spatio-temporal data image corresponding to a vehicle 200. In this manner the velocity of said corresponding vehicle 200 is determined more accurately.
[0104] The present invention also provides a system 100 for extraction of characteristic parameters of traffic data patterns from traffic. The traffic comprises at least one moving vehicle 200 on a road. The system 100 comprises a receiver 120 configured for performing the method according to the present invention. The system 100 may further comprise the sensing optical fiber cable 130 and the pulse generating device 110.
[0105] It is contemplated that some of the steps discussed herein as software methods may be implemented within hardware, for example, as circuitry that cooperates with the processor to perform various method steps. Portions of the present invention may be implemented as a computer program product wherein computer instructions, when processed by a computer, adapt the operation of the computer such that the methods and / or techniques of the present invention are invoked or otherwise provided. Instructions for invoking the inventive methods maybe stored in fixed or removable media, transmitted via a data stream in a broadcast or other signal bearing medium, and / or stored within a working memory within a computing device operating according to the instructions.
[0106] Although various embodiments which incorporate the teachings of the present invention have been shown and described in detail herein, those skilled in the art can readily devise many other varied embodiments that still incorporate these teachings.
[0107] A final remark is that embodiments of the present invention are described above in terms of functional blocks. From the functional description of these blocks, given above, it will be apparent for a person skilled in the art of designing electronic devices how embodiments of these blocks can be manufactured with well-known electronic components. A detailed architecture of the contents of the functional blocks hence is not given.
[0108] While the principles of the invention have been described above in connection with specific apparatus, it is to be clearly understood that this description is made only by way of example and not as a limitation on the scope of the invention, as defined in the appended claims.
Claims
Claims1. Computer implemented method for extraction of characteristic parameters of traffic, said traffic comprising at least one moving vehicle (200) on a road, said method comprising the steps of:transmitting, by a pulse generating device (110), an optical signal (310) into a sensing optical fiber cable (130) positioned along said road; and receiving, by a receiver (120), from said sensing optical fiber cable (130) a reflected optical signal (320) resulting from acoustic vibrations in said sensing optical fiber cable (130) caused by said at least one moving vehicle (200); andderiving, by at least one processing unit (122), a spatio-temporal data image d from said reflected optical signal (320),characterised in that said method further comprises the steps, performed by said at least one processing unit (122), of:determining for at least one pixel in said spatio-temporal data image d a gradient-square tensor G;determining a slope p for said at least one pixel of said spatio-temporal data image d from the gradient-square tensor G; andconverting said slope p into a velocity of the corresponding vehicle (200) on the road by means of a linear transformation of said slope p.
2. Computer implemented method according to claim 1 , wherein the gradientsquare tensor G is determined as G = wherein gtt= gxx=3. Computer implemented method according to claim 1 or 2, wherein the slope p is determined from the largest eigenvalue of the gradient-square tensor G.
4. Computer implemented method according to any one of the claims 1 -3, at least in combination with claim 2, wherein the slope p is determined as p9tx whereinmaxis the largest eigenvalue of the gradient-square tensor G.
5. Computer implemented method according to any one of the claims 1-4, wherein the step of determining the gradient-square tensor G further comprises:applying a predetermined smoothing filter on the components of the gradient square tensor G.
6. Computer implemented method according to claim 5, at least in combination with claim 2, wherein the step of determining the gradient-square tensor G further comprises:applying said predetermined smoothing filter on gtt, gxxand gtx.
7. Computer implemented method according to any one of the claims 1-6, wherein the method further comprises the steps, performed by the at least one processing unit (122), of:transforming said spatio-temporal data image d into a F-K domain representation; andapplying a predetermined F-K filter on the F-K domain representation configured for filtering out frequencies and / or wavenumbers outside a predetermined velocity range; andinverse transforming said F-K domain representation to the spatiotemporal domain, thereby obtaining the spatio-temporal data image d in velocity filtered form.
8. Computer implemented method according to claim 7, wherein the predetermined velocity range ranges from 10 km / h to 220 km / h, preferably from 12 km / h to 210 km / h, more preferably from 14 km / h to 200 km / h, even more preferably from 16 km / h to 190 km / h, and yet even more preferably from 18 km / h to 180 km / h.
9. Computer implemented method according to any one of claims 1-8, wherein the step of receiving the reflected optical signal (320) further comprises: thresholding, by said at least one processing unit (122), the reflected optical signal (320) to filter out sudden spikes in the reflected optical signal (320).
10. Computer implemented method according to any one of claims 1-9, wherein the method further comprises the step of determining a noise level by:transmitting, by the pulse generating device (110), an optical signal into a reference optical fiber cable; andreceiving, by the receiver (120), from said reference optical fiber cable, a reflected reference optical signal;determining, by said at least one processing unit (122), the noise level from the reflected reference optical signal, andwherein the step of receiving the reflected optical signal (320) further comprises:subtracting, by said at least one processing unit (122), the determined noise level from the reflected optical signal (320).
11. Computer implemented method according to any one of claims 1 -10, wherein the step of receiving the reflected optical signal (320) further comprises: converting, by said at least one processing unit (122), the reflected optical signal (320) to nanostrain by means of a look-up table specific to the pulse generating device (110), the sensing optical fiber cable (130) and the receiver (120).
12. Computer implemented method according to any one of claims 1-11, wherein the step of receiving the reflected optical signal (320) further comprises: applying, by said at least one processing unit (122), a predetermined bandpass filter to the reflected optical signal (320) for filtering out frequencies outside of a predetermined frequency range.
13. Computer implemented method according to claim 12, wherein the predetermined frequency range ranges from 0,1 Hz to 4,0 Hz, preferably from 0,2 Hz to 3,5 Hz, more preferably from 0,3 Hz to 3,0 Hz, even more preferably from 0,4 Hz to 2,5 Hz, and yet even more preferably from 0,5 Hz to 2,0 Hz.
14. Computer implemented method according to any one of claims 1-13, wherein the step of determining the slope p further comprises:applying an error correction on the slope p based on a slope error retrieved from synthetic examples with a known slope.
15. System (100) for extraction of characteristic parameters of traffic, said traffic comprising least one moving vehicle (200) on a road, said system (100) comprising:a sensing optical fiber cable (130) positioned along said road;a pulse generating device (110) configured for transmitting an optical signal (310) into said sensing optical fiber cable (130); anda receiver (120) configured for receiving from said sensing optical fiber cable (130) a reflected optical signal (320) resulting from acoustic vibrations in said sensing optical fiber cable (130) caused by said at least one moving vehicle (200); andat least one processing unit (122) configured for:deriving a spatio-temporal data image d from said reflected optical signal (320);determining for at least one pixel in said spatio-temporal data image d a gradient-square tensor G;determining a slope p for said at least one pixel of said spatiotemporal data image d from the gradient-square tensor G; and converting said slope p into a velocity of the corresponding vehicle (200) on the road by means of a linear transformation of said slope p.
16. Computing device for a system (100) for extraction of characteristic parameters of traffic, said traffic comprising least one moving vehicle (200) on a road, said system (100) comprising:a sensing optical fiber cable (130) positioned along said road;a pulse generating device (110) configured for transmitting an optical signal (310) into said sensing optical fiber cable (130); anda receiver (120) configured for receiving from said sensing optical fiber cable (130) a reflected optical signal (320) resulting from acousticvibrations in said sensing optical fiber cable (130) caused by said at least one moving vehicle (200),wherein said computing device comprises at least one processing unit (122) configured for:deriving a spatio-temporal data image d from said reflected optical signal (320);determining for at least one pixel in said spatio-temporal data image d a gradient-square tensor G;determining a slope p for said at least one pixel of said spatio-temporal data image d from the gradient-square tensor G; andconverting said slope p into a velocity of the corresponding vehicle (200) on the road by means of a linear transformation of said slope p.