CN109269403B

CN109269403B - Contact net lead abrasion measuring method and device based on machine vision

Info

Publication number: CN109269403B
Application number: CN201710579763.XA
Authority: CN
Inventors: 李想; 彭博; 曹伟; 王瑞锋
Original assignee: Chengdu Tang Source Electrical Ltd By Share Ltd
Current assignee: Chengdu Tang Source Electrical Ltd By Share Ltd
Priority date: 2017-07-17
Filing date: 2017-07-17
Publication date: 2020-02-04
Anticipated expiration: 2037-07-17
Also published as: CN109269403A

Abstract

The invention relates to the field of measurement of rigid wires of subway system contact networks in urban rail transit. Aiming at the problems in the prior art, the invention provides a method and a device for measuring the abrasion of a contact net lead based on machine vision. Wear measurement is achieved by a basic method that utilizes image processing and machine vision. In the collected image containing the conducting wire, through image processing thresholding and connected region characteristic analysis, L1, L2 and L3 are divided from the rest of the background in the image, the conducting wire L2 is determined by utilizing the conducting wire boundary positions E1 and E2, and the width of L2 is measured to obtain the abrasion value d. Furthermore, if the wear value d of the wire at each position of the wire L2 cannot be accurately calculated because the wire is only discontinuous in a local area in an image due to the complexity of a subway line, the accurate position of the wire is obtained by processing such as wire tracking and image thinning, and the measured wire wear value d is calculated.

Description

Contact net lead abrasion measuring method and device based on machine vision

Technical Field

The invention relates to the field of measurement of rigid conductors of subway system overhead contact systems in urban rail transit, in particular to a method and a device for measuring abrasion of conductors of the overhead contact systems based on machine vision.

Background

In an electrified railway, a train operates to obtain current through mutual contact sliding of a pantograph slide plate and a contact network wire. In order to ensure that the pantograph receives current well, a certain contact pressure is required between the pantograph sliding plate and the contact line. However, with the vibrations caused by train operation and the long-term contact between the pantograph slide and the contact wire, the contact wire will have different degrees and different angles of wear.

From the knowledge of physics, the wear of the contact line is mainly related to the pressure of the pantograph, the characteristics of the pantograph slide plate and the state of the contact line surface. The rigid suspension busbar has no elasticity, is used for offsetting the action force of the pantograph lifting force, and suddenly appears along with the occurrence of the pantograph lifting force on the contact suspension, so that the impact of the pantograph lifting force on the contact suspension is larger, and the abrasion degree of a contact line is larger. According to the related technical indexes of railway lines, when the contact line is abraded to a certain degree, maintenance or replacement is needed, otherwise, safety accidents are easily caused. Therefore, there is a need for irregular measurement monitoring of contact line wear conditions of subway lines.

Disclosure of Invention

The technical problem to be solved by the invention is as follows: aiming at the problems in the prior art, the method and the device for measuring the abrasion of the contact net lead based on the machine vision are provided. Wear measurement is achieved by a basic method that utilizes image processing and machine vision. In the collected image containing the conducting wire, through image processing thresholding and connected region characteristic analysis, L1, L2 and L3 are separated from the rest parts, the conducting wire L2 is determined by utilizing the positions of the conducting wire boundary regions E1 and E2, and the width of L2 is measured, so that the abrasion value d can be obtained. Further, if the wear value d cannot be accurately calculated, the accurate position of the wire is obtained by adopting the processing of wire tracking, image thinning and the like, so that the measured wire wear value d is calculated.

The technical scheme adopted by the invention is as follows:

a contact net wire abrasion measuring method based on machine vision comprises the following steps:

an image acquisition step: the method comprises the steps that a single-path camera collects an image of a contact network wire at the bottom of the contact network wire;

and a step of conducting wire maximum communication area: separating a maximum connection area of the contact wires including L1, L2 and L3, a boundary position E1 and a boundary position E2 from the contact wire image;

and a lead positioning step: obtaining a thresholded image T (x, y) based on the thresholding processing of the maximum connected region of the lead; according to the characteristics of the connected regions, the thresholded image T (x, y) is combined with the boundary position E1 and the boundary position E2 to obtain position information of L1, L2 and L3;

and a wear value calculation step: the wire wear value d was calculated using the positional information of L2 using the principle of image processing.

Further, when the wire wear value d cannot be calculated, after the wire positioning step, before the wire wear value calculation step, a wire profile tracking step is further included, specifically:

refining the L1, L2, and L3 center positions;

on the basis of the maximum connected domain of the lead, the center positions of L1, L2 and L3 are combined with the boundary position E1 and the boundary position E2, and the discontinuity of the leads of L1, L2 and L3 is identified;

and tracking and identifying discontinuous parts of L1, L2 and L3 wires in the image by utilizing the gradient of the maximum connected domain of the wires in the vertical direction and the continuity of gradient gray distribution to obtain the complete contour position of L1, L2 or L3 in the image.

Further, refining the center positions of L1, L2 and L3 specifically means:

on the basis of obtaining L1, L2 and L3 and bus boundaries E1 and E2 through calculation, thinning the maximum connected regions where L1, L2 and L3 are located by using a binary image thinning algorithm, and thus obtaining 3 center lines corresponding to L1, L2 and L3.

Further, the specific process of dividing the maximum connection region of the lead, the boundary position E1 and the boundary position E2 is as follows: by setting an empirical threshold, the maximum connection region of the wire including L1, L2, and L3 is distinguished from the background region, resulting in the maximum connection region of the wire including L1, L2, and L3.

Further, the thresholding of the maximum connected region of the lead is to acquire a gradient image of the original image through a computing camera and perform thresholding of the maximum connected region of the lead based on a threshold T of the gradient image to obtain a thresholded image T (x, y).

Further, the gradient image threshold t is calculated as follows:

acquiring gradient image values E (x, y) of an image I (x, y) at the bottom of a rigid wire of the overhead line system according to the single-path camera, and calculating gradient image values of the image of the overhead line based on gray value weighting

And obtaining t by the ratio of the image value of the contact line image gradient.

Furthermore, the single-path camera is installed on the top of the detection train, is positioned below a contact network wire and is responsible for acquiring contact network wire images including the bottom of the contact network wire.

Further, the wire abrasion value d is calculated by calculating the gray scale change of the lateral direction of the position of the wire L2 by using the position information of the wire L2 and adopting an image processing method to obtain the abrasion width of the wire L2, and the wire abrasion value d is obtained according to the width of the wire L2.

A contact net wire wear measuring device based on machine vision includes:

the image acquisition module is used for acquiring an image of the contact network lead at the bottom of the contact network lead through the single-path camera;

a wire maximum connection area module for dividing a wire maximum connection area containing L1, L2 and L3, a boundary position E1 and a boundary position E2 from the contact network wire image;

and the wire positioning module is used for performing thresholding processing on the maximum connected region of the wire to obtain a thresholded image T (x, y), and combining the boundary position E1 and the boundary position E2 according to the characteristics of the wire connected region to obtain the position information of L1, L2 and L3.

And the abrasion value calculation module is used for calculating the wire abrasion value d by utilizing the position information of the L2 and adopting the principle of image processing.

In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that:

the linear array camera is adopted to realize the acquisition of the abrasion image of the contact wire lead, so that the abrasion condition of the lead is easier to monitor visually;

the method comprises the following steps of carrying out binarization processing on a contact wire abrasion image twice, namely, firstly, segmenting a maximum connected region of a wire comprising L1, L2 and L3 from a background, and realizing the binary processing by using a lower empirical threshold (extracting the maximum connected region of the whole wire comprising L1, L2 and L3 by a method of setting the empirical threshold according to general gray level distribution of line wire imaging); and the gradient map value of the maximum communication area of the lead is binarized through the accurately calculated t value for the second time, and L1, L2 and L3 are accurately segmented with the rest images. Under an ideal state, the width of the lead L2 can be obtained, and the purpose of accurate measurement is achieved; the core positioning method of the contact net lead is based on the gradient image, and can overcome various interferences caused by complex and various subway line environments to a great extent.

In addition to the implementation process, the scheme also has a corresponding processing process for special situations, namely when the wire wear value d cannot be calculated, judging the L1, the L2 and the L3 images, and further judging the condition that the wires of the L1, the L2 and the L3 are discontinuous, so that the measurement accuracy of the method is improved.

The wear measurement of the contact line conductor is completed by adopting a non-contact method, so that the real-time wear state detection of the whole line can be realized, and the timely feedback of maintenance and replacement is facilitated;

the method can meet the abrasion measurement of different pull-out value range standards of the contact network lead, and is easy to expand.

Drawings

The invention will now be described, by way of example, with reference to the accompanying drawings, in which:

figure 1 is a schematic representation of a rigid conductor of a catenary.

Figure 2 is a comparison graph of worn wire.

Fig. 3 is a rigid wire diagram of a catenary.

Fig. 4 is a general algorithm flow chart.

Fig. 5 is a flow chart relating to lead positioning.

Fig. 6 is a wire wear detection flowchart.

Detailed Description

All of the features disclosed in this specification, or all of the steps in any method or process so disclosed, may be combined in any combination, except combinations of features and/or steps that are mutually exclusive.

Any feature disclosed in this specification may be replaced by alternative features serving equivalent or similar purposes, unless expressly stated otherwise. That is, unless expressly stated otherwise, each feature is only an example of a generic series of equivalent or similar features.

Description of the background:

at present, the domestic subway line contact net mainly comprises a rigid contact net. The two major parts of the rigid catenary core are a busbar and a contact wire, as shown in fig. 1, and herein, unless otherwise specified, the catenary wire refers to the complete rigid contact wire including two components of the busbar and the contact wire in fig. 1, while the wire refers to the contact wire in fig. 1, and the busbar refers to the busbar in fig. 1. In the running process of the train, the current taking is realized by the sliding contact between the pantograph slide plate on the top of the train and the lead; the contact net wire of same circuit will appear the wire wearing and tearing condition gradually through the slip of different train pantograph slide plates for a long time, simultaneously because the contact pressure between different pantographs and the wire is different, will lead to the wearing and tearing state inconsistent of different contact line positions, consequently is necessary to measure the monitoring to circuit contact net wire wearing and tearing state to overhaul the maintenance to the circuit as early as possible.

As shown in fig. 2, the sectional view of the new unworn lead and the worn lead is clearly distinguished, the lower half portion of the new lead is in a complete arc shape, the bottom end of the worn lead is in a plane shape, the wear width is w along the image in the transverse direction, and the wear measurement needs to complete the measurement of the wear width w of L2. The final data presentation mode of wearing and tearing is different according to different demands, and the final wearing and tearing measurement data value of this scheme presents with the data form of tangent plane wearing and tearing diameter length d.

Fig. 3 is a basic outline view of a rigid conductor of an overhead line system, wherein L1 represents a left busbar end of the overhead line system, L2 represents a section part with a wear width w formed by abrasion of the rigid conductor when a pantograph pan is actually contacted with the overhead line system, and L3 represents a right busbar end of the overhead line system. The boundary position E1 and the boundary position E2 respectively correspond to the left boundary and the right boundary of the wire L1 and the L3 in the image acquired by the single-path camera and including the bottom of the rigid wire of the overhead line system. The camera is arranged on the roof of the detection train and is responsible for acquiring images of contact network wires, and as the L1, the L2 and the L3 are closer to the camera than the background part, the L1, the L2 and the L3 can be well distinguished from background area images by setting empirical values in the acquired images; the wire abrasion plane forms an image which needs to be identified at two ends of the wire L2 and the busbars L1 and L3 with abrasion width w (three grid line areas of L1, L2 and L3 in FIG. 3), and the rest is a background area (a white area in the middle of L1 and L3). x represents the coordinate of the direction parallel to the ground in the contact line image, and y represents the coordinate of the direction perpendicular to the x-axis in the contact line image. In this patent, the catenary wire is primarily made up of these three components. In the actual camera captured image, L1, L2, and L3 are in lighter strip-like areas, and the background area is darker. Generally, the widths of the two ends L1 and L3 of the bus bar are fixed, the width between L1 and L3 is also fixed, and the section of the unworn lead is smaller than the section of the worn lead. x represents a lateral direction coordinate parallel to the ground in the contact line image, and y represents a longitudinal direction coordinate perpendicular to the x-axis direction in the contact line image.

The wire maximum communication area segmentation module, the contact network wire segmentation module and the wire abrasion value calculation module are all realized by hardware equipment such as a processor with a data processing function.

The working principle of the invention is as follows:

the abrasion measurement of the contact network lead is completed by a method of positioning the contact network lead in a camera acquisition image by utilizing image processing and analyzing the gray distribution characteristics of the lead. Firstly, a camera finishes the acquisition of a conductor imaging image within an effective pull-out value range, and then an image processing algorithm is utilized to segment the maximum communication area of the rigid conductor of the overhead line system including L1, L2 and L3 and the rest parts of the rigid conductor of the overhead line system so as to judge the image acquisition effectiveness of the camera for abrasion measurement; finally, in an effective camera acquisition image, the maximum communication area of the rigid wire of the overhead line system comprising L1, L2 and L3 is identified through gray imaging gray characteristic and edge gradient characteristic analysis of the wire, and the wire abrasion width w is detected according to the wire abrasion width d of L2.

The first embodiment is as follows:

step 1: the method comprises the steps that a single-path camera collects an image of the overhead contact system at the bottom of an overhead contact system rigid lead;

step 2: dividing a contact net lead maximum communication area comprising L1, L2 and L3, a boundary position E1 and a boundary position E2 from the contact net image;

the specific process of dividing the maximum connection area, the boundary position E1 and the boundary position E2 of the lead in the step 2 is as follows: according to the general gray distribution characteristics of the contact network wire imaging, the maximum connection area of the wire including L1, L2 and L3 is distinguished from the background area by setting an empirical threshold, and the maximum connection area of the wire including L1, L2 and L3 is obtained.

And step 3: carrying out thresholding processing on the maximum connected region of the contact network wire to obtain a thresholded image T (x, y), combining a boundary position E1 and a boundary position E2 according to the characteristics of the maximum connected region of the contact network wire, and segmenting L1, L2 and L3 and the rest background images of the maximum connected region of the wire to obtain position information of L1, L2 and L3, wherein the position information refers to left and right boundaries and width information of L1, L2 and L3;

the thresholding treatment of the maximum connected region of the contact network wire in the step 3 is to acquire a gradient image value of a contact network image through a calculation camera and carry out thresholding treatment on the maximum connected region of the wire based on a gradient image threshold t;

the specific process of thresholding the maximum connected region of the rigid conductor of the contact network in the step 3 to obtain a thresholded image T (x, y) is as follows:

step 31, adopting a gradient operator to calculate and extract a wire gradient image, which is beneficial to removing most of noise in the image, and adopting a gradient operator template:

a single-path camera is arranged to acquire a junction contact line image comprising the bottom of a rigid lead of a contact network, namely an image I (x, y) acquired by an original camera, and a gradient image value E (x, y) is equal to E (I (x, y));

and step 32, carrying out thresholding processing on the gradient image value E through a threshold value T to obtain a thresholded image T (x, y).

Wherein the t calculation process is as follows: acquiring gradient image values E (x, y) of an image I (x, y) at the bottom of a rigid wire of the overhead line system according to the single-path camera, and calculating gradient image values of the image of the overhead line based on gray value weightingAnd obtaining t by the ratio of the image value of the contact line image gradient. The specific process is as follows:

let the gradient image be E (x, y), then the threshold value for thresholding the gradient image

Wherein: e.g. of the type_x＝E_x(I(x,y))，e_y＝E_y(I(x,y))，e_xy＝(e_x+e_y)·I(x,y)，S_xy＝∑e_xy，

Wherein-e_xA gradient image value indicating a horizontal direction of a certain pixel coordinate x (current pixel) in the contact line image; e.g. of the type_yA gradient image value which represents the vertical direction of a certain pixel coordinate y (current pixel) of the contact line image; e.g. of the type_xyRepresenting the total gradient value of a pixel point at a certain pixel coordinate (x, y) (current) in the contact line image; s_xyRepresenting a contact line image gradient image value sum;

representing the sum of contact line image gradient image values weighted based on gray scale values.

And 4, step 4: the wire wear value d is calculated using the principle of image processing using the positioning information of the wire L2.

In step 4, specifically, the positioning information of the lead L2 is utilized, an image processing method is adopted, the wear width w of the lead L2 is calculated through the gray scale change of the transverse position of the lead L2, and the lead wear value d is obtained according to the wear width w of the lead L2. And (4) setting the radius of the wire as r according to the conversion of the geometrical relation of the circle. Thus, the wire abrasion value d:

d ═ Γ (w, r); Γ is the geometrical transformation relation between the wire abrasion value d and w, r.

Example two: on the basis of the first embodiment, when the L2 width information cannot be identified by one image capturing camera and the wire wear value d cannot be calculated, it is further determined whether the single image is discontinuous due to L1 and L2, i.e., L3 wires in the image; the method comprises the following specific steps:

and 5: when the wire abrasion value d can not be calculated, the method further comprises the following steps:

step 51: refining the L1, L2, and L3 center positions; on the basis of obtaining L1, L2 and L3 and bus boundaries E1 and E2 through calculation, thinning the maximum connected regions where L1, L2 and L3 are located by using a binary image thinning algorithm, and obtaining 3 center lines corresponding to L1, L2 and L3;

step 52: on the basis of the maximum connected domain of the lead, the center positions of L1, L2 and L3 are combined with the left and right boundaries E1 and E2 of the lead, and discontinuity of the leads of L1, L2 and L3 is identified; l1, L2, and L3 wire discontinuities refer to any of the following three ways:

1) any one of the conductors L1, L2 and L3 is discontinuous;

2) any two of L1, L2, and L3;

3) the three conductors L1, L2 and L3 are not continuous at the same time;

step 53: due to the complex subway line environment and the difference of collected images and the like, the positions of the L1, L2 and L3 components obtained in step 52 are incomplete, and then step 53 calculates the minimum gradient change in two adjacent lines of images of the vertical direction conductor by using the continuity of the gradient and the gray distribution of the maximum connected domain of the conductor in the vertical direction, tracks and identifies discontinuous partial images of the L1, L2 and L3 conductors in the images, and obtains the complete contour positions of L1, L2 or L3 in the images.

And 54, smoothing the positions of the L1, the L2 and the L3 based on the gray level change. This step is primarily to make the curve profile positions of L1, L2, and L3 smoother.

Wherein, the step 51 of refining the center positions of L1, L2 and L3 specifically means that: based on the positions of L1, L2 and L3 and the boundary positions E1 and E2, the method is completed according to a thinning algorithm, the center positions of L1, L2 and L3 are obtained, and 3 corresponding center lines of L1, L2 and L3 are formed.

The binary image thinning algorithm generally refers to an operation of skeletonization of a binary image, and is a short term of a process of reducing lines of an image from a multi-pixel width to a unit pixel width. At present, the image algorithm of the direction is mature and applied more, such as algorithms of Hilditch, Pavlidis, Rosenfenld and the like. The widely used Hilditch refinement algorithm is employed herein.

The Hilditch refinement algorithm comprises the following steps:

assume a background value of 0 and a foreground value of 1. The 3x3 neighborhood structure for pixel p is:

x4	x3	x2
			x5	p	x1
x6	x7	x8

each pixel is iterated from left to right and top to bottom of the image as an iteration cycle. In each iteration cycle, for each pixel p, it is marked if it simultaneously satisfies the following 6 conditions. At the end of the current iteration cycle, the values of all marked pixels are set to background values. If no marker point (i.e., a pixel satisfying 6 conditions) is present in an iteration cycle, the algorithm ends.

The 6 conditions are:

(1): p is 1, i.e. p is not background;

(2): x1, x3, x5 and x7 are not all 1 (otherwise, the p mark is deleted and the image is empty);

(3): at least 2 of x 1-x 8 are 1 (if only 1 is 1, it is the end point of the line segment, if not, it is the isolated point);

(4): the number of 8-linked linkages of p is 1;

the join number refers to the number of graph components connected to p in the 3x3 neighborhood of pixel p:

(5) assuming that x3 has been marked for deletion, when x3 is 0, the 8-way ligation number for p is 1;

(6) assuming x5 has been marked for deletion, when x5 is 0, the 8-way ligation number for p is 1.

The specific process of obtaining the complete contour positions of L1, L2 and L3 in the image in step 53 is as follows: through step 52, the positions of L1, L2, and L3 in the image can be calculated, but the positions of L1, L2, and L3 are not continuous; and (3) continuously calculating gradient values of positions of broken (discontinuous) lines and effective line wires in a certain curve such as L1 at the longitudinal broken (discontinuous) position, and tracking by calculating gradient minimum values to obtain the positions of broken lines, so as to obtain the complete contour position by filling.

The invention is not limited to the foregoing embodiments. The invention extends to any novel feature or any novel combination of features disclosed in this specification and any novel method or process steps or any novel combination of features disclosed.

Claims

1. A contact net wire abrasion measuring method based on machine vision is characterized by comprising the following steps:

and a step of conducting wire maximum communication area: extracting a maximum connection area of the contact wires including L1, L2 and L3, a boundary position E1 and a boundary position E2 from the contact wire image; wherein, L1 represents the left busbar end of the contact line wire, L2 represents the tangent plane part with abrasion width formed by abrasion of the rigid wire when the pantograph slide plate is actually contacted with the contact line wire, and L3 represents the right busbar end of the contact line wire; the boundary position E1 and the boundary position E2 correspond to the L1 left boundary and the L3 right boundary respectively;

and a lead positioning step: obtaining a thresholded image T (x, y) based on the thresholding processing of the maximum connected region of the lead; according to the characteristics of the connected regions, the thresholded image T (x, y) is combined with the boundary position E1 and the boundary position E2 to obtain position information of L1, L2 and L3; wherein the position information refers to left and right boundaries and width information of L1, L2, and L3; the thresholding of the maximum connected region of the lead is to acquire a gradient image of an original image through a calculation camera and carry out thresholding of the maximum connected region of the lead based on a gradient image threshold T to obtain a thresholded image T (x, y);

2. The method for measuring the abrasion of the contact net conductor based on the machine vision according to the claim 1, characterized in that when the abrasion value d of the conductor cannot be calculated, after the conductor positioning step, before the abrasion value calculation step, a conductor profile tracking step is further included, specifically:

refining the L1, L2, and L3 center positions;

and tracking and identifying discontinuous partial images of L1, L2 and L3 leads in the image by utilizing the continuity of the gray scale and gradient distribution in the vertical direction of the maximum connected domain of the leads to obtain the complete contour position of L1, L2 or L3 in the image.

3. The method for measuring the contact net wire abrasion based on the machine vision as claimed in claim 2, wherein the refining of the center positions of L1, L2 and L3 specifically means that:

4. The method for measuring the abrasion of the contact net conductor based on the machine vision as claimed in claim 2, wherein the maximum connection area of the conductor, the boundary position E1 and the boundary position E2 are divided, and the specific process is as follows: by setting an empirical threshold, the maximum connection region of the wire including L1, L2, and L3 is distinguished from the background region, resulting in the maximum connection region of the wire including L1, L2, and L3.

5. The contact net wire abrasion measuring method based on the machine vision according to the claim 1, characterized in that the gradient image threshold t calculation process is:

6. The contact net lead abrasion measuring method based on the machine vision as claimed in claim 1, characterized in that the camera is installed on the top of the detection train, is positioned under the contact net lead, and collects the contact net lead image including the bottom of the contact net lead.

7. The method as claimed in claim 1, wherein the wire wear value d is calculated by calculating the gray scale change of the wire L2 in the transverse direction to obtain the wear width of the wire L2 and the wire wear value d is obtained from the L2 width by using the position information of the wire L2 and by using an image processing method.

8. The contact net wire wear measuring device based on the machine vision contact net wire wear measuring method of one of claims 1 to 6, characterized by comprising:

a wire maximum connection region extraction module, configured to divide a wire maximum connection region including L1, L2, and L3, a boundary position E1, and a boundary position E2 from the catenary wire image; wherein, L1 represents the left busbar end of the contact line wire, L2 represents the tangent plane part with abrasion width formed by abrasion of the rigid wire when the pantograph slide plate is actually contacted with the contact line wire, and L3 represents the right busbar end of the contact line wire; the boundary position E1 and the boundary position E2 correspond to the L1 left boundary and the L3 right boundary respectively;

the wire positioning module is used for carrying out thresholding processing on the maximum connected region of the wire to obtain a thresholded image T (x, y), and combining the boundary position E1 and the boundary position E2 according to the characteristics of the wire connected region to obtain the position information of L1, L2 and L3; wherein the position information refers to left and right boundaries and width information of L1, L2, and L3; the thresholding of the maximum connected region of the lead is to acquire a gradient image of an original image through a calculation camera and carry out thresholding of the maximum connected region of the lead based on a gradient image threshold T to obtain a thresholded image T (x, y);