CN103217111A - Non-contact contact line geometrical parameter detecting method - Google Patents

Abstract

The invention discloses a non-contact contact line geometric parameter detection method. It includes the following steps: firstly, collect high-definition images at equal time intervals by collecting control signals, and then use median filtering, image grayscale and other technologies to complete image preprocessing; then use threshold iteration method and digital morphology to remove isolated noise to realize Positioning and coordinate extraction of the center point of the laser spot; then extract the matching target area, and detect its horizontal gray scale singular value; then use the "image coordinate system-camera coordinate system" and "camera coordinate system-detection vehicle coordinate system" Convert, give the wire height and pull-out value of the contact wire at this place, and then compensate the vibration of the car body; finally give the accurate detection value of the guide height and pull-out value, and display several parameter information on the developed graphical monitoring interface middle. The method of the invention effectively improves the detection efficiency of the catenary geometric parameters, simplifies the algorithm, improves the accuracy of fault detection, and can specifically improve the safety and reliability of the high-speed railway catenary.

Description

A non-contact contact line geometric parameter detection method

technical field

The invention relates to the field of high-speed railway detection, in particular to a novel non-contact contact line geometric parameter detection method based on image processing. the

Background technique

In order to prolong the service life of the locomotive pantograph on electrified railways and make the sliding plate of the pantograph wear evenly, the contact wires are arranged in a zigzag shape in the straight line area of the line. If the catenary pull-out value is set small, the purpose of uniform sliding wear and prolonging the life of the pantograph cannot be achieved; if it is set too large, in case of bad weather, the contact line will exceed the limit, resulting in bow scraping or bow drilling accidents, except Because of the looseness of hardware parts and tilting of pillars, etc., the pull-out value will also exceed the design value. The zigzag pull-out value is generally (480±10) mm, so the pull-out value of the catenary must be checked frequently. In order to make good contact between the electric locomotive pantograph and the overhead contact wire, and to ensure proper contact force, there is a certain requirement for the height of the contact wire, which is (6000±30) mm. the

At present, the detection methods used for the geometric parameters of catenary at home and abroad mainly include: falling line plus steel ruler, road ruler measurement method; measuring rod plus special calculator measurement method; optical measurement method; DJJ multifunctional catenary detector, etc. These detection methods have achieved certain results, but many methods have problems such as inaccurate measurement, high risk, complicated operation, expensive and cumbersome equipment, heavy detection tasks, and poor anti-interference ability. The detection of geometric parameters of contact lines based on image processing has not been reported yet. the

Contents of the invention

The technical problem to be solved by the present invention is to provide a new non-contact contact line geometric parameter detection method based on image processing to solve the existing shortcomings of low contact line geometric parameter detection efficiency and poor system flexibility. the

Technical solution of the present invention is:

A non-contact contact line geometric parameter detection method, which collects high-definition images in the geometric parameter detection device configured in this way: the laser emits linear laser light to illuminate the contact line directly above it, and then passes through the high-resolution image sensor installed on the roof of the detection vehicle. High frame rate camera device acquisition, in which the camera and laser are placed on the horizontal beam on the inspection vehicle, the camera and the beam present a certain angle, the laser is perpendicular to the beam and faces upward, and the beam is directly below the catenary and parallel to the center line of the track; A vibration sensor is installed in the detection device to realize the monitoring of the vibration waveform; to complete the correction of the geometric parameters of the contact line, the precise detection value of the conduction height and the pull-out value of the catenary contact line is obtained through image processing, including the following steps:

A. Image acquisition step, the laser emission line laser hits the contact line to present bright spots, and the CCD camera acquisition device is used to collect high-definition images of the contact line in real time;

B, image data processing step, carry out preprocessing to the collected image, and realize the detection and positioning of the position where the laser hits the contact line, find out the coordinates of the center point position of the laser spot in the imaging plane coordinate system, and complete the position in " The conversion of "image coordinate system-camera coordinate system" and "camera coordinate system-detection vehicle coordinate system" gives the wire height and pull-out value of the contact line at this place;

C. According to the vibration signal of the vibration sensor, detect the vibration waveform after the car has driven for many times, fit the vibration waveform, and compensate and correct the above-mentioned wire height and pull-out value;

D. Displaying the monitoring information in a graphical monitoring interface. the

In the image data processing step, since the position where the line laser strikes the contact line has the highest intensity and the gray value of this position in the image is the largest, the image is directly binarized, and then the optimal threshold value iteration and digital morphology are used to remove the isolated The noise method is used to complete the positioning of the laser spot and the extraction of the coordinates of the center point of the spot in the imaging plane coordinate system. the

In the image data processing step, the coordinates of the central point of the laser spot are also included in the conversion of the "image coordinate system-camera coordinate system" and "camera coordinate system-detection vehicle coordinate system", so as to realize the height of the wire at this position, Detection of pull values. the

The graphical monitoring interface includes: contact wire guide height monitoring controls, contact wire pull-out value monitoring controls, vibration detection controls, and also includes storage of guide height normal information and fault information. the

The visual acquisition and measurement method of the present invention is used to analyze the contact line captured by the high-definition camera, and to process massive images in real time under the operation of the inspection vehicle. Specifically, a laser and a high-resolution high-frame-rate CCD camera are installed on the roof beam of the test vehicle. The laser emits line laser light to the contact line, collects running images of the contact line at equal intervals, and completes the alignment of the contact line within the set interval. Calculation of guide height and pull-out value. Then, through the detection model training and online testing, the correction scheme is adapted to the measurement of the geometric parameters of the contact line of the detected line. the

The novel non-contact contact line geometric parameter detection method based on image processing proposed by the present invention mainly has the following advantages:

1. The present invention detects the geometric parameters of the catenary directly through the image processing method. The algorithm is simple and convenient, and a large number of images can be processed in batches. good means. the

2. Machine vision systems have become more cost-effective as computer processor prices have dropped dramatically compared to optical methods. the

3. Compared with optical sensors, machine vision systems have better flexibility and adaptability, and their hardware equipment is relatively fixed. All that is needed is a corresponding change or upgrade of the software rather than the addition of expensive hardware equipment. the

4. The operation and maintenance costs of the machine vision system are very low, and the system is stable. the

5. The detection method based on image processing is non-contact, without any wear and danger to catenary components, and can obtain more accurate guide height and pull-out value data. the

Therefore, the technical solution proposed by the present invention is a very suitable and promising detection means for the new non-contact contact line geometric parameter detection based on image processing. This technology can perform non-contact and online measurement on the contact wire during the operation of the inspection vehicle, and overcomes the shortcomings of the traditional method that can only work under static conditions and is difficult to adapt to the needs of high-speed railways. Detection performance, and theoretically have the ability of high-speed online measurement, has a good application prospect. the

Description of drawings

Fig. 1 is a schematic diagram of the detection device of the present invention. the

FIG. 2 is a configuration diagram of a hardware device in an embodiment of the present invention. the

Fig. 3 is a schematic diagram of the idea of the method proposed in this paper of the present invention. the

Fig. 4 is a diagram showing the parameters of the inspection vehicle according to the embodiment of the present invention. the

Fig. 5 is a schematic diagram of a computer image of the present invention. the

FIG. 6 is a schematic diagram of a triangle similarity relationship used in the present invention. the

Fig. 7 is a flowchart of laser spot positioning in the present invention. the

Fig. 8 is the image laser spot location figure of the present invention (Fig. 8a: original picture, Fig. 8b: median filtering effect diagram, Fig. 8c: shape

Figure 8d: Laser spot positioning map). the

Fig. 9 is a schematic diagram of interface functions of the present invention. the

Detailed ways:

The embodiment of the present invention will be described in further detail below in conjunction with the actual pictures taken in the experiment. the

As shown in FIG. 1 , it is a schematic diagram of the detection device of the present invention. The beam of the detection vehicle is located directly above the center line of the track; the vertical beam of the laser is facing upwards, and the emitted laser area is larger than the maximum pull-out value of the contact line; the CCD camera and the beam present a fixed angle. the

The working principle is: when the inspection vehicle is moving along the track, the laser emission line hits the contact line to form a brighter spot, and the CCD camera located at the rear end of the inspection vehicle will take pictures according to the set time interval. . With the difference of the height and position of the contact line, the position of the laser spot in the image has a corresponding presentation. By locating the position of the light spot, the conduction height of the contact line can be better reflected, and the value can be pulled out. Under the premise of not considering the height change of the rail surface, the distance between the light spot and the bottom of the image can reflect the guide height, and the distance between the light spot and the center of the image can reflect the change of the pull-out value. the

FIG. 2 is a configuration diagram of a hardware device in an embodiment of the present invention. The hardware equipment of the present invention is mainly composed of the camera device on the cross beam of the detection vehicle, the laser light source device for measurement, the vibration detection device at the bottom of the vehicle, the data acquisition and processing unit (notebook computer) for image acquisition and processing, and the vehicle body vibration compensation measurement at the bottom of the vehicle. device. the

Fig. 3 is a schematic diagram of the idea of the method proposed in this paper of the present invention. The main steps are: including the following steps: first, use the acquisition control signal to collect high-definition images at equal time intervals, and then use median filtering, image grayscale and other technologies to complete image preprocessing; then use threshold iteration method and digital morphology to remove isolated noise The method realizes the positioning and coordinate extraction of the center point of the laser spot; then extracts the matched target area and detects its horizontal gray scale singular value; and then uses the "image coordinate system - camera coordinate system" and "camera coordinate system - detection vehicle Coordinate system" conversion, give the wire height and pull-out value of the contact line at this place, and then compensate the vibration of the car body; finally give the accurate detection value of the guide height and pull-out value, and display several parameter information on the developed In the graphical monitoring interface. the

Fig. 4 is a diagram showing the parameters of the inspection vehicle according to the embodiment of the present invention. The main optical axis of the camera is a straight line l, the horizontal distance between the laser and the camera is L, the plane where the image is located is AB, the distance between the beam of the detection vehicle and the contact line is H ₁ , and the vertical distance between the beam of the detection vehicle and the center line of the rail is d.

As shown in Figure 4, the pinhole model of the CCD camera is built in the embodiment as follows:

The models used in CCD camera calibration are generally divided into two types: linear and nonlinear. The linear model is the pinhole model. The calibration method used in the present invention detects the vehicle coordinate system, and the definition of the camera coordinate system and the computer coordinate system is shown in FIG. 4 . the

The origin of the coordinate system of the inspection vehicle is at the center of gravity of the inspection vehicle, the x-axis is perpendicular to the paper surface, the y-axis is perpendicular to the rail surface, and the z-axis is along the track to the right; the origin of the camera coordinate system is o _a , z _a and the main optical axis of the camera Keep the same direction, y _a is vertical to z _a upwards, x _a axis is vertical to the inside of the paper, consistent with the coordinate system of the inspection vehicle. In the figure, the imaging plane is AB, perpendicular to the optical axis, and the distance o _a is f (the focal length of the camera).

It is easy to know that the camera coordinate system rotates θ clockwise, then moves -d along the y-axis of the rotated coordinate system, and moves L/2 along the z-axis, which is the detection vehicle coordinate system. the

The rotation transformation homogeneous matrix is:

$\mathrm{<span\; class="notranslate">\; Rot</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& \mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]$

The translation transformation homogeneous matrix is:

$\mathrm{<span\; class="notranslate">\; Trans</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; d</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]$

Obviously, the homogeneous matrix of the detection vehicle coordinate system relative to the camera coordinate system is:

$\mathrm{<span\; class="notranslate">\; T</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; Rot</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; Trans</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; d</span>}}_{\mathrm{<span\; class="notranslate">\; 1</span>}}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left[\begin{array}{cccc}\mathrm{<span\; class="notranslate">\; 1</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; LG</span>}\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& \mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}& <span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; d</span>}\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; LG</span>}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; 1</span>}\end{array}\right]$

Due to the particularity of the installation of the contact wire, the erection presents a zigzag feature, and the response in the image is shown in Figure 5. x represents the pull-out value at the location of the laser point on the contact line, and y represents the conductance at the laser point. the

Linear models are solved according to linear equations, which are simple and fast. In this paper, the small hole imaging model is used for calibration. According to the relationship of similar triangles, as shown in Figure 6, it can be directly obtained:

y=f _y , gy ₁ /D

Similarly: x＝f _x gx ₁ /D

The point coordinates of the detection vehicle coordinate system corresponding to (x, y) in the image are (x _r , y _r , z _r ), then:

$\left\{\begin{array}{c}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; Lx</span>}<span\; class="notranslate">\; /</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;\; g</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}<span\; class="notranslate">\; )</span>\\ {\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; yL</span>}\mathrm{<span\; class="notranslate">\; cos</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}}{\mathrm{<span\; class="notranslate">\; sin</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}{\mathrm{<span\; class="notranslate">\; gf</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; the\; tan</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; yL</span>}}{\mathrm{<span\; class="notranslate">\; the\; tan</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}{\mathrm{<span\; class="notranslate">\; gf</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; L</span>}\mathrm{<span\; class="notranslate">\; the\; tan</span>}\mathrm{<span\; class="notranslate">\; \&theta;</span>}\\ {\mathrm{<span\; class="notranslate">\; z</span>}}_{\mathrm{<span\; class="notranslate">\; r</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}\end{array}\right.$

Fig. 7 is a flowchart of laser spot positioning in the present invention. The iterative threshold method is used in the positioning of the center point of the laser spot. This method is an excellent method in the image segmentation of the threshold method. The algorithm first selects an approximate threshold as the initial value of the estimated value, and then performs image segmentation to generate sub-images, and according to the image Select a new threshold based on the characteristics of the algorithm, and then use the new threshold to segment the image. After several cycles, the wrongly segmented pixels can be reduced to the minimum. the

The implementation steps of the optimal threshold iterative algorithm in Figure 7 are as follows:

(a) Assuming that there is no information about the exact position of the object, the first approximation is to set the average gray value of the image as the initial threshold. The rationality of this approach is reflected in the fact that the average grayscale must be between the background grayscale and the object grayscale;

(b) using a threshold, the images are divided into two groups, _Q1 and _Q2 respectively;

(c) calculate the gray mean values R ₁ and R ₂ of Q ₁ and Q ₂ ;

(d) Find a new threshold T _n = (R ₁ +R ₂ )/2;

(e) If T _n+1 =T _n , then end the iteration, otherwise proceed to step b.

Finally, take T _n+1 as the optimal segmentation threshold. Generally, the algorithm can converge to the optimal threshold after about 5 iterations, which has a certain degree of self-adaptability.

窗口中心点为f_i。再将这些点按数值大小排序，取序号中心点数值作为滤波输出。  The image median filtering process in Fig. 7 is realized as follows: the median filtering is to use a sliding window with odd points, and replace the value of the center point of the window with the median value of each point in the window. Assume the sequence f ₁ , f _{2 ,} L f _n . Take the window length as m (m is an odd number), and perform median filter processing, that is, extract m numbers f _iv , L f _i-1 , f _i , f _i+1 , L,

The center point of the window is f _i . Then sort these points according to the numerical value, and take the value of the center point of the serial number as the filter output.

${\mathrm{<span\; class="notranslate">\; the\; y</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; Med</span>}<span\; class="notranslate">\; \{</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; L</span>}{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; L</span>}<span\; class="notranslate">\; ,</span>{\mathrm{<span\; class="notranslate">\; f</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; v</span>}}<span\; class="notranslate">\; \}</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; \&Element;</span>\mathrm{<span\; class="notranslate">\; Z</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; v</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; 1</span>}}{\mathrm{<span\; class="notranslate">\; 2</span>}}$

Median filtering has invariance, can overcome the loss of details and blurred image edges caused by linear filters such as least mean square filtering and average filtering, and has a better filtering effect on salt and pepper noise in images. the

In Figure 7, the implementation of digital morphology to remove isolated noise is as follows:

The mathematical morphology method is used for processing. First corrode and then expand, in order to avoid too many holes inside the moving target, conditional corrosion is adopted:

(1) If the current pixel A(x ₀ ,y ₀ ) is white, traverse its 3×3 neighborhood.

(2) If a point B(x ₁ ,y ₁ ) is black, go to (3); otherwise go to (4).

(3) Traverse the 7×7 neighborhood of point A, count the number of white points in the neighborhood, if it is less than the threshold T, corrode point A, that is, assign the value to black. the

(4) Process the next pixel. the

The image after corrosion processing left some small voids, so the image was expanded by 3×3. From Figure 8, it can be seen that the internal voids at the laser spot basically disappeared after the expansion process. the

The image after denoising still has some noise points due to camera shake and other factors. Due to the large area, the morphological method cannot be eliminated. Here, the hole detection is performed on the image, and then each hole is checked, which is a connected white area. If the number of internal white points is less than the threshold T, all pixels in the current hole will be assigned a background value of 0 for elimination. The large white area left in the final image is the laser spot target to be extracted. the

Claims (5)

1. contactless osculatory geometric parameter detection method, gather high-definition image in the geometric parameter pick-up unit with formation like this: laser instrument is launched the osculatory of linear illuminated with laser light directly over it, the high frame per second camera head of high resolving power collection by being installed on the inspection vehicle roof again, wherein video camera and laser instrument are placed on the horizontal gird on the inspection vehicle, video camera and crossbeam present certain angle, laser instrument perpendicular to crossbeam up, crossbeam under contact net and with the orbit centre line parallel; The monitoring of vibration transducer realization to vibrational waveform is installed in the described pick-up unit; The correction of finishing contact line geometry parameter has the accurate detected value of leading height, stagger that obtains the contact net osculatory through Flame Image Process, may further comprise the steps:

A, image acquisition step, laser instrument emission line laser is beaten and present speck on osculatory, utilizes the ccd video camera harvester, gathers the osculatory high-definition image in real time;

B, view data treatment step, the image of gathering is carried out pre-service, and realize laser is wherein beaten detection and location in the osculatory position, obtain the coordinate of laser facula center position in the imaging plane coordinate system, finish of the conversion of this position, provide lead height and the stagger of osculatory at this place at " image coordinate system-camera coordinate system " and " camera coordinate system-inspection vehicle coordinate system ";

C, according to the vibration signal of vibration transducer, the vibrational waveform after inspection vehicle repeatedly travels, the match vibrational waveform compensates above-mentioned lead height and stagger, revises;

D, described monitor message is presented in the graphical monitoring interface.

2. the method for claim 1, it is characterized in that, in the described view data treatment step, because line laser is beaten position intensity maximum on osculatory, should locate the gray-scale value maximum in the image, directly image binaryzation is handled, utilized optimal threshold iteration and digital morphological to learn the method for removing isolated noise again, finish the location of laser facula and the extraction of spot center point coordinate in the imaging plane coordinate system.

3. the method for claim 1, it is characterized in that, in the described view data treatment step, also comprise of the conversion of the coordinate of laser facula center position at " image coordinate system-camera coordinate system " and " camera coordinate system-inspection vehicle coordinate system ", thereby realize this position lead height, the detection of stagger.

4. the method for claim 1 is characterized in that, described graphical monitoring interface comprises: osculatory is led high monitoring control, osculatory stagger monitoring control, vibration detection control.

5. the method for claim 1 is characterized in that, described graphical monitoring interface also comprises and storing leading high normal information and failure message.

Images