Fiber-optic moving target monitoring method and system based on image processing and clustering techniques

Abstract

The application discloses a kind of based on image processing and clustering technique's optical fiber moving target monitoring method and system, the present application utilizes the vibration signal that distributed optical fiber is collected using image processing technology and density peak clustering algorithm is handled, the trajectory of moving target is extracted, to monitor the running condition of vehicle on road, calculate traffic flow, traffic speed and other traffic parameters, to generate reverse driving alarm, overspeed alarm, fault early warning, road condition early warning and other alarm early warning information and moving target quantity, average speed, average vehicle distance and other statistical information of travel, and these information can be mapped to actual road distribution diagram, show map and vehicle real-time information are pushed to user terminal, can effectively track vehicle trajectory, monitor the running condition of vehicle and the congestion situation of road.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of distributed optical fiber detection technology, and particularly relates to a method and system for monitoring moving targets of optical fibers based on image processing and clustering technology. BACKGROUND

[0002] In recent years, with the development of economy, the country has formed a relatively developed highway network through the great construction of the state. The rapid development of highways provides important support for the healthy operation of the economy and the overall development of society. However, on the other hand, with the increasing demand for transportation, various traffic incidents also occur from time to time, such as road damage, traffic accidents, cargo scattering, and vehicle anchoring, which can cause road traffic congestion, and traffic congestion often leads to more traffic hazard events. Traffic incidents cause a large amount of delay, and also cause environmental pollution, which seriously affects people's travel experience and reduces the efficiency of social travel, which is a common problem faced by countries around the world.

[0003] With the rapid development of highway construction in China, the number of tunnels in highways is also increasing rapidly. Tunnels generally have the special nature of "mountain crossing" and are relatively closed with relatively poor visibility. Compared with ordinary highways, the sudden event of tunnels has a high casualty rate and heavy loss. Not only are accidents prone to occur, but secondary accidents are also likely to occur. In order to ensure that various environments, road conditions, and accident information in the tunnel can be fed back to the vehicle driver and the tunnel management station in a timely manner, a safety warning system will be set up in the tunnel. Through various collection devices, information in the tunnel is collected and published in a timely manner, greatly improving the safety of tunnel driving.

[0004] No matter whether it is an ordinary highway, a highway, or a tunnel, timely collection of traffic information, real-time measurement of traffic flow and speed, and other traffic parameters can help discover traffic bottlenecks, detect traffic abnormalities in a timely manner, reduce traffic accidents, reduce energy consumption, and reduce environmental pollution.

[0005] Currently, there are many traffic information collection technologies in practical application, such as the coil detection technology, which is the first road contact type traffic information collection technology developed, and the video detection technology and radar detection technology, which are newly developed road non-contact traffic information collection technologies. There are also other ultrasonic detection technologies, infrared detection technologies, etc.

[0006] The traffic information collected through these technologies mainly includes vehicle speed, traffic flow, lane occupancy rate, and vehicle spacing. When there is vehicle overspeed, tunnel parking, slow vehicle driving, tunnel congestion, or traffic accidents, the corresponding measures are triggered in a timely manner according to the emergency plan prepared in advance.

[0007] There are still many problems in current road condition information collection technology, such as that the vehicle detector in the tunnel is generally arranged at the entrance and exit of the tunnel, and the actual vehicle running track in the tunnel cannot be judged; the video detection technology is realized through high-definition cameras and rear-end algorithms, but the coverage is insufficient, and it cannot work effectively at night or when the visibility in the tunnel is low.

[0008] The current traffic information collection technology is mainly a point type local area monitoring method, which is not suitable for wide range and full distribution monitoring, and is easily affected by extreme weather. SUMMARY

[0009] Therefore, the present application provides a kind of optical fiber moving target monitoring method and system based on image processing and clustering technology, the trajectory of moving target is extracted by using image processing technology and density peak clustering algorithm, to monitor the running condition of vehicle on road, calculate traffic flow, vehicle speed and other traffic parameters.

[0010] A kind of optical fiber moving target monitoring method based on image processing and clustering technology, specifically comprising the following steps:

[0011] S1, the vibration signal on road is collected by using the distributed optical fiber buried and laid, and the strip trajectory waterfall of moving target is obtained;

[0012] S2, the contour area of strip trajectory waterfall is subjected to Hough transform, and a large number of straight line segments are obtained;

[0013] S3, the density peak clustering algorithm is used, the interference line segment is filtered out according to the descriptor of all straight line segments, and the cluster number of remaining useful line segments and the trajectory of each cluster corresponding to moving target are obtained;

[0014] S4, the slope deviation degree between the trajectory of each cluster and each trajectory in trajectory library and the intercept deviation degree of x axis are calculated, and the trajectory of the same moving target is matched from trajectory library;

[0015] S5, according to the trajectory of each cluster and the trajectory matched from trajectory library corresponding thereto, the moving speed and spacing of moving target are calculated;

[0016] S6, the trajectory, moving speed, spacing and moving state of moving target are updated to trajectory library;

[0017] S7, the alarm and early warning information and statistical information of moving target in daily monitoring are generated according to the trajectory, moving speed, spacing and moving state of moving target, and all information is reported to server, server maps the received information to actual road distribution map according to the set road setting, and finally the display map and vehicle real-time information are pushed to the terminal of user.

[0018] Preferably, the specific steps in step S3 for filtering out interfering line segments based on the descriptors of all line segments are as follows:

[0019] Construct the descriptor Q for all line segments. i ;

[0020] Based on the descriptor Q of all line segments i The relative distance δ between line segments is calculated using the density peak clustering algorithm. i and local density ρ i ;

[0021] Based on the calculated relative distance δ i and local density ρ i Interference segments are filtered out; the interference segments are those where δ < δ thr or ρ < ρ thr The straight line segment of the region.

[0022] Preferably, the descriptor Q of the straight line segment i =(k i , l i -b i / k i ), where k i Let l be the slope parameter of the line segment. i Let b be the length of the line segment. i Let b be the y-intercept of the line segment. i / k i It is the absolute value of the x-intercept of the intersection point of the extension of the line segment and the x-coordinate.

[0023] Preferably, the specific steps for obtaining the number of clusters of remaining useful line segments and the trajectory of the moving target corresponding to each cluster in step S3 are as follows:

[0024] Select any one line segment from the useful line segments and initialize it as the first cluster center. Calculate the distances from the remaining line segments to this cluster center. Line segments within a distance threshold range belong to the same cluster.

[0025] Select any line segment from the line segments outside the distance threshold range and add it as the second cluster center. Calculate the distance values ​​from the remaining line segments to the cluster center in turn. Find the line segments that belong to the same cluster as the original line segment from the remaining line segments. Repeat this step until all line segments have been calculated.

[0026] The mean of the line segments in each cluster is taken as the new cluster center C. m ;

[0027] Calculate the descriptor Q for each line segment in each cluster. i With its cluster center C mthe distance between the two points, if the calculated distance value is less than a threshold value Q thr , then the straight line segment belongs to the core cluster member of its cluster;

[0028] For each cluster, the average value of the slope parameter of its core cluster member and the average value of the intercept on the y-axis are taken as the trajectory of the moving target corresponding to the cluster;

[0029] For each cluster, the maximum and minimum values of the x-axis of the straight line segment endpoints of its core cluster member are calculated and the maximum and minimum values of the y-axis are calculated

[0030] Preferably, the specific step of matching the trajectory of each cluster to the trajectory of the same moving target in the trajectory library in step S4 is as follows:

[0031] The number of pixels H of the downward movement of the strip trajectory waterfall diagram under the fixed identification period is calculated;

[0032] The slope deviation degree K m_n between the trajectory of the mth cluster in the current frame and the nth trajectory in the trajectory library is calculated;

[0033] The intercept deviation degree P m_n on the x-axis between the trajectory of the mth cluster in the current frame and the nth trajectory in the trajectory library is calculated;

[0034] According to the slope deviation degree K m_n and the intercept deviation degree P m_n , it is determined whether the trajectory of the mth cluster and the nth trajectory in the trajectory library belong to the trajectory of the same moving target, if P m_n <P Thr and K m_n <K thr , then the nth trajectory in the trajectory library and the trajectory of the mth cluster are the trajectory of the same moving target.

[0035] Preferably, the calculation step of the intercept deviation degree P m_n on the x-axis between the trajectory of the mth cluster in the current frame and the nth trajectory in the trajectory library is as follows:

[0036] The intercept d n of the nth trajectory in the trajectory library on the x-axis is calculated, and the calculation formula of the intercept d n is as follows:

[0037]

[0038] According to the intercept d n of the nth trajectory in the trajectory library on the x-axis, the intercept of the nth trajectory on the x-axis in the current frame is calculated intercept The calculation formula is:

[0039]

[0040] Calculate the x-intercept d of the trajectory of the m-th cluster in the current frame. m , intercept d m The calculation formula is:

[0041]

[0042] Calculate the deviation P between the intercept of the m-th trajectory in the current frame and the n-th trajectory in the trajectory library. m_n Deviation P m_n The calculation formula is:

[0043]

[0044] Preferably, the calculation steps for calculating the moving target's speed and distance in step S5 are as follows:

[0045] Based on the trajectory of the m-th cluster and its corresponding m'-th trajectory matched from the trajectory database, calculate the movement distance D of the moving target corresponding to the m-th cluster. m Distance D m The calculation formula is:

[0046]

[0047] Then the moving speed v of the moving target corresponding to the m-th cluster m The calculation formula is:

[0048]

[0049] Among them, D m The unit is meters, and T is the recognition period, which is in milliseconds.

[0050] Calculate the distance R between the two moving targets before and after the current recognition period. m The calculation formula is:

[0051]

[0052] Calculate the average distance between all moving targets in the current identification period. The calculation formula is:

[0053]

[0054] Preferably, the specific steps in step S6 for updating the trajectory, speed, spacing, and movement status of the moving target to the trajectory database are as follows:

[0055] If the mth cluster matches a trajectory of a same moving target in the trajectory library, it is judged whether the minimum value of the y-axis of the end point of the straight line segment of the core cluster member of the mth cluster satisfies If yes, it indicates that the trajectory of the straight line segment has been interrupted, and the moving target has stopped moving, so the moving state of the moving target is updated to stop; if It indicates that the trajectory of the straight line segment continues, and the moving target is moving, so the real-time moving speed v m of the moving target is updated The moving state of the moving target is updated to moving; if The current position of the moving target is updated to If The current position of the moving target is updated to

[0056] If the mth cluster does not match a trajectory of a same moving target in the trajectory library and there is no target stopping moving nearby, the moving target corresponding to the cluster is marked as a new moving target, and the start position and the current position of the new moving target are updated and created in the trajectory library; when The current position is The start position is When The current position is The start position is

[0057] A fiber moving target monitoring system based on image processing and clustering technology, comprising:

[0058] A distributed optical fiber is used for collecting vibration signals on a road and transmitting the collected data to an upper computer;

[0059] The upper computer is used for processing the received data based on image processing technology and a density peak clustering algorithm, obtaining the trajectory, moving speed, interval and moving state of a moving target, generating alarm and early warning information and statistical information of the moving target in daily monitoring, and reporting all information to a server;

[0060] The server is used for mapping the received information to an actual road distribution map according to a set road setting, and finally pushing a display map and real-time vehicle information to a user terminal;

[0061] And the user terminal.

[0062] The beneficial effects of the present application are:

[0063] 1, The application utilizes image processing technology and density peak clustering algorithm to process the vibration signals collected by the distributed optical fiber, extracts the trajectory of the moving target, monitors the running status of the vehicle on the road, calculates the traffic parameters such as vehicle flow and vehicle speed, generates warning and early warning information such as reverse driving warning, overspeed warning, fault early warning and road condition early warning, and statistical information such as the number of moving targets, the average speed of driving and the average distance of driving, and can map these information to the actual road distribution map, push the display map and real-time vehicle information to the user terminal, which can effectively track the vehicle trajectory, monitor the running status of the vehicle and the congestion of the road.

[0064] 2, The application utilizes distributed optical fiber to collect vibration signals on the road, and the distributed optical fiber has the following advantages:

[0065] 1) Not affected by bad weather.

[0066] 2) Not affected by light illumination, etc.

[0067] 3) Long distance monitoring can be performed, and the road in a range of tens of kilometers can be continuously and real-timely monitored.

[0068] 4) Each position in the whole process can be monitored, and the scheme combined with a traffic camera can capture and take evidence of the violation phenomenon such as overspeed. Only one camera needs to be set at the intersection to monitor the whole road.

[0069] 5) The trajectory of the moving target can be real-timely tracked, so the moving target such as vehicle can be real-timely tracked and recorded in the high-security area.

[0070] 6) The movement of personnel, construction vehicles in mine, pipe gallery and other areas can be tracked, and the action of personnel inspection and maintenance can be monitored. BRIEF DESCRIPTION OF DRAWINGS

[0071] In order to more clearly illustrate the technical solutions of the embodiments of the application, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0072] Figure 1 It is a schematic diagram of a strip trajectory waterfall chart.

[0073] Figure 2 It is a schematic diagram of moving targets A and B concentrated in two clusters.

[0074] Figure 3 It is a schematic diagram of matching the moving target in the trajectory library.

[0075] Figure 4 This is the flow chart of the method of the present invention.

[0076] Figure 5 This is the composition diagram of the system of the present invention.

[0077] Figure 6 This is a diagram showing an actual application of the system of the present invention in vehicle monitoring in a park.

[0078] The meanings of the reference numerals in the figure are as follows:

[0079] 1 is an interference line segment, 2 is a useful line segment, 3 is the current frame, and 4 is the previous frame (track library). Specific embodiments

[0080] For a better understanding of the technical solution of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0081] It should be clear that the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0082] The following will further describe the present application in detail through specific embodiments in combination with the accompanying drawings.

[0083] The present invention provides a fiber optic moving target monitoring method based on image processing and clustering technology, which specifically includes the following steps:

[0084] S1. Use a distributed optical fiber laid underground to collect vibration signals on the road and obtain a strip trajectory waterfall diagram of the moving target.

[0085] S2. Perform a Hough transform on the contour area of the strip trajectory waterfall diagram to obtain a large number of straight line segments.

[0086] As Figure 1 shown, Figure 1 in addition to the moving trajectory of the real moving target (the useful line segment 2 described below), there may also be some interference line segments 1.

[0087] After performing a Hough transform on the contour area of the strip trajectory waterfall diagram, a large number of straight line segments can be obtained. Each straight line segment is expressed by a vector, that is, each straight line segment is represented by a vector (x1, y1, x2, y2) with four elements, where (x1, y1) and (x2, y2) are the two endpoints of each straight line segment.

[0088] S3. Use the density peak clustering algorithm to filter out the interference line segments according to the descriptors of all the straight line segments and obtain the number of clusters of the remaining useful line segments and the trajectories of the moving targets corresponding to each cluster.

[0089] Specifically, the steps to filter out interfering line segments are as follows:

[0090] S31, Create the descriptor Q for all line segments. i .

[0091] In this embodiment, the slope-intercept form of the linear equation is used to determine the model of each line segment, and the slope parameter of the line segment is used.

[0092] The y-intercept of the line segment is b = y1 - k × x1;

[0093] Length of a straight line segment

[0094] The x-intercept of the intersection point of the extension of the straight line segment and the x-coordinate is -b / k.

[0095] For each line segment i obtained by the Hough transform, it can be represented as:

[0096] Y i =k i X i +b i

[0097] Therefore, the descriptor Q of the line segment i =(k i , l i -b i / k i ).

[0098] S32, based on the descriptor Q of all line segments i The relative distance δ between line segments is calculated using the density peak clustering algorithm. i and local density ρ i .

[0099] S33, based on the calculated relative distance δ i and local density ρ i Interference segments are filtered out; the interference segments are those where δ < δ thr or ρ < ρ thr The straight line segment of the region.

[0100] like Figure 2 As shown, Figure 2 There are two moving targets, A and B. Each moving target has multiple straight line segments. After using the density peak clustering algorithm, the trajectories of moving targets A and B will be concentrated into two clusters, while other interfering straight line segments are distributed in the region close to the horizontal or vertical axis. These interfering line segments have too low density or are too close to the density center to form a new density center.

[0101] The density peak value of the useful line segments after filtering needs to satisfy the following two conditions simultaneously: the local density p is greater than a set threshold value, and the relative distance d is greater than a set threshold value. That is

[0102] d > d thr and p > p thr

[0103] The specific steps for obtaining the number of clusters of the remaining useful line segments and the trajectory of the moving target corresponding to each cluster are as follows:

[0104] S34, based on the principle that the line segments of the same belt trajectory are close in distance and the line segments of different belt trajectories are far in distance, the cluster center C m and the number of clusters M are calculated. The specific method is as follows:

[0105] An arbitrary line segment is selected from the useful line segments, and the line segment is initialized as the first cluster center C1. The distance values of the remaining line segments to the cluster center are calculated in turn, and the line segments within the distance threshold range belong to a cluster.

[0106] Then, an arbitrary line segment is selected from the line segments outside the distance threshold range, and the line segment is added as the second cluster center. The distance values of the remaining line segments to the cluster center are calculated in turn, and the line segments belonging to the same cluster as the line segment are found from the remaining line segments. The step is repeated until all line segments are calculated.

[0107] The mean value of the line segments in each cluster is taken as the new cluster center C m of the cluster.

[0108] Alternatively, an arbitrary line segment is selected from the useful line segments, and the line segment is initialized as the first cluster center C1. M is set to 1. Then, the distance values of the remaining line segments to the cluster center are calculated in turn. If the calculated distance value is within the set distance threshold range, it is considered that the corresponding line segment belongs to the same cluster as the first line segment, and the mean value of the line segments in the cluster is taken as the new cluster center. Otherwise, the line segment not within the distance threshold range is added as the second cluster center C2, the number of clusters M is set to M+1, and the distance of the remaining line segments to the first cluster center C1 and the second cluster center C2 is calculated. Whether it belongs to the first cluster or the second cluster or a new cluster is determined according to the calculated distance value. The step is repeated until all line segments are calculated.

[0109] S35, the distance between the descriptor Q i of each line segment in each cluster and the descriptor Q m of its cluster center C m is calculated. If the calculated distance value is less than the threshold Q thr , the line segment belongs to the core cluster member of its cluster.

[0110] S36, for each cluster, average the slope parameter k of all core cluster members and the average of the intercept b on the y-axis as the trajectory of the moving target corresponding to the cluster;

[0111] for each cluster, calculate the maximum and minimum of the x-axis of the line segment endpoints of all core cluster members and the maximum and minimum of the y-axis and the maximum and minimum

[0112] S4, calculate the slope deviation and the intercept deviation on the x-axis between the trajectory of each cluster and each trajectory in the trajectory library, and match the trajectories of the same moving target from the trajectory library.

[0113] Suppose there are N trajectories in the trajectory library, and M clusters are identified in the current recognition period, that is, M moving trajectories. For each line segment trajectory m (m = 0, 1, … M) of the current frame, match each trajectory n (n = 0, 1, …, N) in the trajectory library to find trajectories belonging to the same moving target.

[0114] As shown in Figure 3 , specifically, the steps of matching the trajectories of the same moving target from the trajectory library for each cluster are:

[0115] S41, calculate the number of pixels H of the downward movement of the strip trajectory waterfall under the fixed recognition period;

[0116]

[0117] Calculate the slope deviation K m_n between the trajectory of the mth cluster of the current frame and the nth trajectory in the trajectory library.

[0118]

[0119] Calculate the intercept d n of the nth trajectory in the trajectory library on the x-axis.

[0120]

[0121] According to the intercept d n of the nth trajectory in the trajectory library on the x-axis, calculate the intercept of the current frame on the x-axis

[0122]

[0123] Calculate the intercept d m of the trajectory of the mth cluster of the current frame on the x-axis.

[0124]

[0125] Calculate the deviation P of the intercept of the mth track of the current frame and the nth track in the track library m_n ;

[0126]

[0127] S42, according to the slope deviation K m_n And the intercept deviation P m_n Determine whether the track of the mth cluster and the nth track in the track library belong to the same track of the moving target, if P m_n <P Thr And K m_n <K thr , the nth track in the track library and the mth track of the moving target are the same track.

[0128] S5, according to the track of each cluster and the corresponding track matched from the track library, the moving speed and distance of the moving target are calculated.

[0129] When calculating the moving speed of the moving target, based on two principles:

[0130] 1. In a short time window, the speed of the moving target is constant;

[0131] 2. The shape of the end of the strip-shaped track is irregular, and the position deviation is large when the end of the strip-shaped track is calculated. The intercept of the track in the x axis is basically at the center position of the track strip, and the calculation position is more reliable.

[0132] Specifically, the steps of calculating the moving speed and distance of the moving target are:

[0133] S51, according to the track of the mth cluster and the corresponding m'th track matched from the track library, the cross section of the track in the x axis is used to calculate the moving distance D m of the moving target corresponding to the mth cluster. m The calculation formula of the moving distance D

[0134]

[0135] The calculation formula of the moving speed v m of the moving target corresponding to the mth cluster is:

[0136]

[0137] Where, D m is in meters, and T is the identification period, in milliseconds.

[0138] Considering that the speed of the moving target changes fast sometimes, the average speed is calculated by smoothing the moving speed

[0139]

[0140] S52, the distance R between the front and rear moving targets in the current identification period is calculated m The calculation formula is:

[0141]

[0142] The average distance between all moving targets in the current identification period is calculated The calculation formula is:

[0143]

[0144] S6, the trajectory of the moving target is updated and the moving speed, the distance and the moving state are updated to the trajectory library.

[0145] Specifically, if the mthcluster matches the trajectory of the same moving target in the trajectory library, i.e. satisfies

[0146]

[0147] whether the minimum value of the y-axis of the end point of the straight line segment of the core cluster member of the mthcluster satisfies If yes, it indicates that the trajectory of the straight line segment has been interrupted, the moving target has stopped moving, and the moving state of the moving target is updated to stop;

[0148] If it indicates that the trajectory of the straight line segment continues, the moving target is moving, and the real-time moving speed v m of the moving target is updated, the average moving speed and the moving state of the moving target is updated to moving;

[0149] If the current position of the moving target is updated as

[0150] If the current position of the moving target is updated as

[0151] If the mthcluster does not match the trajectory of the same moving target in the trajectory library and there is no target stopping moving nearby, the moving target corresponding to the cluster is marked as a new moving target, and the start position and the current position of the new moving target are updated and created in the trajectory library; when the current position is The start position is When , the current position is The start position is

[0152] If the mth cluster is not matched to the trajectory of the same mobile target in the trajectory library and there is a target with a "stop" state nearby, the trajectory library is updated and marked as "restart".

[0153] S7, according to the trajectory, moving speed, distance and moving state of the mobile target, generate the alarm and early warning information and statistical information of the mobile target in daily monitoring, and report all the information to the server. The server maps the received information to the actual road distribution map according to the set road setting, and finally pushes the display map and vehicle real-time information to the user's terminal.

[0154] Specifically, the alarm and early warning information includes reverse driving alarm, overspeed alarm, fault early warning, road condition early warning, etc.

[0155] When the driving trajectory of the mobile target is opposite to the direction of the normal driving trajectory, a reverse driving alarm is generated and sent out. In this embodiment, the parameter of road direction is set according to the location of the host. If the host is on the starting side of the monitoring road, the road direction parameter is set as K base =1, and on the termination side, it is set as K base =-1. If It is explained that the driving trajectory of the vehicle is opposite to the direction of the normal driving trajectory, the ID information, position information, driving trajectory information and time information of the mobile target are recorded, and a reverse driving alarm is generated.

[0156] When the average speed of the mobile target exceeds the set threshold value , it is considered that the speed of the mobile target is overspeed, the ID information, position information, driving trajectory information and time information of the mobile target are recorded, and an overspeed alarm is generated.

[0157] If more than one mobile target appears in the "stop" state in a short time and in a close position, it is considered that an abnormal condition occurs, such as road abnormality, mobile target accident, etc. The ID information, position information, driving trajectory information and time information of the mobile target are recorded, and a fault early warning is generated.

[0158] If the average speed of multiple mobile targets continuously appears in a position is less than the set threshold value , it is considered that there may be an obstacle or danger affecting traffic in the position, the ID information, position information, driving trajectory information and time information of the mobile target are recorded, and a road condition early warning is generated.

[0159] The statistical information includes the number of moving targets, the average speed of driving, the average distance of driving, etc.

[0160] The application also provides a fiber moving target monitoring system based on image processing and clustering technology, comprising a distributed optical fiber, an upper computer, a server and a user terminal.

[0161] The distributed optical fiber is used for collecting vibration signals on the road and transmitting the collected data to the upper computer.

[0162] The upper computer is used for processing the received data based on the image processing technology and the density peak clustering algorithm, obtaining the trajectory, moving speed, interval and moving state of the moving target, generating the alarm and early warning information and the statistical information of the moving target in daily monitoring, and reporting all the information to the server.

[0163] The server is used for mapping the received information to the actual road distribution map according to the set road setting, and finally pushing the display map and the real-time vehicle information to the user terminal.

[0164] In order to illustrate the effect, the specific embodiments of the application are used to track and monitor the vehicles entering the park to verify the effectiveness of the above method and system. In the engineering project, the optical cable is directly buried in the underground of the road, the window identified by the above method is 50 pixels, the frame length is 15 seconds, and the identification period is once per second. The lower computer (optical cable) collects the vibration data of the road vehicles and transmits the data to the upper computer. The upper computer processes the data by algorithm, and finally reports the result information, including vehicle trajectory identification, vehicle ID allocation, vehicle state, vehicle position, vehicle speed and vehicle number, to the server. The server maps the position information to the actual road distribution map according to the set park road setting, and pushes the display map and the real-time road information to the PC end and the mobile phone end of the user by using the HTML5+CSS+JavaScrip mode.

[0165] The application also provides an electronic device, comprising at least one processor and a memory connected with the processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned method.

[0166] The electronic device is intended to represent various forms of digital electronic computing devices, such as desktop computers, servers, blade servers, mainframe computers, etc. The electronic device can also represent various forms of mobile devices, such as smart phones, wearable devices, etc. The components, their connections and relationships, and their functions described in the application are only for example, and are not intended to limit the implementation of the disclosure described and / or claimed herein.

[0167] It should be understood that the processes shown above can be altered, re-ordered, added to, or deleted from without departing from the scope of the present disclosure. For example, the recited steps of the present disclosure can be performed in parallel, in series, or in a different order, as long as the desired results of the present disclosure are achieved. The present disclosure is not limited in this regard.

[0168] The above description is merely illustrative of the application, and not restrictive. Since modifications, changes, revisions, and improvements can readily occur to those skilled in the art, it is the applicant's intention that the application be limited only as by the scope of the following claims.

Claims

1. A method for monitoring moving targets based on image processing and clustering techniques, characterized in that, Specifically comprising the following steps: S1, collecting vibration signals on the road by using the buried distributed optical fiber to obtain a band trajectory waterfall of the moving target; S2, performing Hough transform on the contour region of the band trajectory waterfall to obtain a large number of straight line segments; S3, filtering out interference line segments according to the descriptors of all straight line segments by using a density peak clustering algorithm, and obtaining the number of clusters of the remaining useful line segments and the trajectory of the moving target corresponding to each cluster; S4, calculating the slope deviation and x-axis intercept deviation between the trajectory of each cluster and each trajectory in the trajectory library, and matching the trajectories of the same moving target from the trajectory library; S5, calculating the moving speed and distance of the moving target according to the trajectory of each cluster and the trajectory matched from the trajectory library; S6, updating the trajectory, moving speed, distance and moving state of the moving target to the trajectory library; S7, generating the alarm and early warning information and statistical information of the moving target in daily monitoring according to the trajectory, moving speed, distance and moving state of the moving target, and reporting all information to the server, and the server maps the received information to the actual road distribution map according to the set road setting, and finally pushes the display map and vehicle real-time information to the user terminal.

2. The method of claim 1, wherein the method is based on image processing and clustering techniques. The specific steps of filtering out interference line segments according to the descriptors of all straight line segments in step S3 are as follows: Establish the descriptor Q of all straight line segments i ; According to the descriptor Q of all straight line segments i , the relative distance δ between straight line segments and the local density ρ are calculated by using the density peak clustering algorithm i i ;​ According to the calculated relative distance δ i and the local density ρ i Filter out interfering line segments, interfering line segments are δ thr or ρ < ρ thr Line segments of the region, δ thr is the relative distance threshold, ρ thr is the local density threshold.

3. The method of claim 2, wherein the method further comprises: Descriptor Q of straight line segment i = (k i , l i , -b i / k i ), wherein k i is a slope parameter of the straight line segment, l i is a length of the straight line segment, b i is a y-axis intercept of the straight line segment, and b i / k i is an absolute value of a cross-intercept of an extension of the straight line segment with an x-coordinate.

4. The method of claim 2, wherein the method further comprises: The specific steps of obtaining the number of clusters of the remaining useful line segments and the trajectory of the moving target corresponding to each cluster in step S3 are as follows: An arbitrary straight line segment is selected from the useful line segments, the straight line segment is initialized as the first cluster center, the distance values of the remaining line segments to the cluster center are calculated in turn, and the line segments within the distance threshold range belong to a cluster, An arbitrary straight line segment is selected from the line segments outside the distance threshold range, the straight line segment is added as the second cluster center, the distance values of the remaining line segments to the cluster center are calculated in turn, and the line segments belonging to the same cluster as the straight line segment are found from the remaining line segments; Repeat this step until all straight line segments are calculated; The mean of the straight line segments in each cluster is taken as the new cluster center C of that cluster m ; Calculate the descriptor Q for each line segment in each cluster i The distance between the line segment and its cluster center C m If the calculated distance value is less than a threshold Q thr Then the line segment is a core cluster member of its cluster; For each cluster, the average of the slope parameters of its core cluster members and the average of the intercepts on the y-axis as the trajectory of the moving target to which the cluster corresponds; For each cluster, calculate the x-axis maximum and minimum of the line segment endpoints of its core cluster members and minimum y-axis maximum and minimum 5. The method of claim 1, wherein the method is characterized by: The specific steps of matching the trajectories of the same moving target from the trajectory library in step S4 are as follows: Calculate the number of pixels H of the band trajectory waterfall moving down under the fixed identification period; calculating a slope deviation degree K between a trajectory of the mth cluster of the current frame and the nth trajectory in the trajectory library m__n ; calculating the intercept deviation P between the x-axis of the trajectory of the mth cluster of the current frame and the nth trajectory in the trajectory library m__n ; According to the slope deviation K m__n With the intercept deviation P m__n If P m__n <P Thr And K m__n <K thr , the n-th trajectory in the trajectory library and the m-th cluster trajectory are the same moving target trajectory.

6. The method of claim 5, wherein the method further comprises: The intercept deviation P between the x-axis of the trajectory of the mth cluster of the current frame and the nth trajectory in the trajectory library m__n The calculation steps are: The intercept d of the nth trajectory in the trajectory library on the x axis is calculated n The calculation formula of the intercept d n is as follows: The intercept d of the nth trajectory in the trajectory library on the x axis n The intercept d of the nth trajectory in the trajectory library on the x axis The intercept d of the nth trajectory in the trajectory library on the x axis The calculation formula is: Calculate the intercept d of the trajectory of the mth cluster in the current frame on the x-axis m The calculation formula of the intercept d m is as follows: calculating a deviation degree P of the mth track of the current frame from the n th track in the track library intercept m__n , Deviation P m__n The formula for calculating the deviation P is: wherein, is the average value of the slope parameter of the core cluster members in each cluster, is the average value of the intercept b on the y-axis of the core cluster members in each cluster, and H is the number of pixels by which the band trajectory waterfall plot is lowered for a fixed identification period.

7. The method of claim 1, wherein the method is based on image processing and clustering techniques. The calculation steps of the moving speed and distance of the moving target in step S5 are as follows: According to the trajectory of the mth cluster and the m'th trajectory matched from the trajectory library corresponding to the m'th cluster, the moving distance D of the moving target corresponding to the mth cluster is calculated m The calculation formula of the moving distance D m is as follows: The moving speed v of the mobile target corresponding to the mth cluster is m The calculation formula is: where D m in meters, T is the identification period in milliseconds. Calculate the distance R between the front and rear moving targets in the current recognition cycle m The calculation formula is: calculating the average distance between all moving targets in the current recognition cycle The formula is: where d m the intercept of the trajectory of the mth cluster of the current frame in the x-axis, is the average of the slope parameters of the core cluster members in each cluster, is the average of the intercepts b of the core cluster members in the y-axis in each cluster.

8. The method of claim 1, wherein the method is based on image processing and clustering techniques. The specific steps of updating the trajectory, moving speed, distance and moving state of the moving target to the trajectory library in step S6 are as follows: If the mth cluster matches a trajectory of the same moving target in the trajectory library, it is determined whether the minimum value of the y-axis of the end point of the straight line segment of the core cluster member of the mth cluster satisfies If yes, it indicates that the trajectory of the straight line segment has been interrupted, the moving target has stopped moving, and the moving state of the moving target is updated to stop; if no it indicates that the trajectory of the straight line segment continues, and the moving target is moving, and the real-time moving speed v m of the moving target is updated and the moving state of the moving target is updated to be moving; if the current position of the moving target is updated to be if the current position of the moving target is updated to be If the mth cluster is not matched to a trajectory of the same moving target in the trajectory library and there is no target stopping moving nearby, the moving target corresponding to the cluster is marked as a new moving target and the starting position and current position of the new moving target are updated in the trajectory library; when the current position is the starting position is the current position is the starting position is the current position is wherein, is the average of the slope parameters for the core cluster members in each cluster.

9. A system for applying the method of monitoring moving targets of optical fibers based on image processing and clustering techniques according to any one of claims 1 to 8, characterized in that, It includes: Distributed optical fiber for collecting vibration signals on the road and transmitting the collected data to the host computer; The host computer is used to process the received data based on image processing technology and density peak clustering algorithm to obtain the trajectory, moving speed, distance and moving state of the moving target, generate the alarm and early warning information and statistical information of the moving target in daily monitoring, and report all information to the server; The server is used to map the received information to the actual road distribution map according to the set road setting, and finally push the display map and vehicle real-time information to the user terminal; And the user terminal.

Images