CN108189859B - Method for judging two laser image characteristics as related redundant characteristics - Google Patents
Method for judging two laser image characteristics as related redundant characteristics Download PDFInfo
- Publication number
- CN108189859B CN108189859B CN201711422597.9A CN201711422597A CN108189859B CN 108189859 B CN108189859 B CN 108189859B CN 201711422597 A CN201711422597 A CN 201711422597A CN 108189859 B CN108189859 B CN 108189859B
- Authority
- CN
- China
- Prior art keywords
- laser image
- laser
- image
- abrasion
- rail
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000012360 testing method Methods 0.000 claims description 8
- 150000001875 compounds Chemical class 0.000 claims description 4
- 238000001514 detection method Methods 0.000 abstract description 58
- 238000005299 abrasion Methods 0.000 description 73
- 229910000831 Steel Inorganic materials 0.000 description 45
- 239000010959 steel Substances 0.000 description 45
- 238000004364 calculation method Methods 0.000 description 10
- 230000007704 transition Effects 0.000 description 8
- 238000012545 processing Methods 0.000 description 7
- 238000007781 pre-processing Methods 0.000 description 6
- 238000005259 measurement Methods 0.000 description 5
- 238000005452 bending Methods 0.000 description 4
- 238000003708 edge detection Methods 0.000 description 4
- 238000000605 extraction Methods 0.000 description 4
- 238000001914 filtration Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 230000003137 locomotive effect Effects 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 210000001367 artery Anatomy 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000010187 selection method Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B61—RAILWAYS
- B61K—AUXILIARY EQUIPMENT SPECIALLY ADAPTED FOR RAILWAYS, NOT OTHERWISE PROVIDED FOR
- B61K9/00—Railway vehicle profile gauges; Detecting or indicating overheating of components; Apparatus on locomotives or cars to indicate bad track sections; General design of track recording vehicles
- B61K9/08—Measuring installations for surveying permanent way
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Length Measuring Devices By Optical Means (AREA)
Abstract
The divisional application discloses a method for judging two laser image features as related redundant features, which belongs to the field of detection, wherein a preferred feature quantity is selected from laser image feature quantities, correlation coefficients of the other laser image feature quantities and the preferred feature quantity are calculated, an average value of the correlation coefficients is calculated, the average value is a threshold β selected by the laser image feature quantities, if the absolute value | rTij | of the correlation coefficient between certain two laser image features is more than or equal to β, the two laser image features are related redundant features, and only one of the two laser image features is selected as the judged laser image feature quantity.
Description
The application is a divisional application with application number 201610765945.1, application date 2016-08-30 and invention name 'automatic detection device for rail abrasion'.
Technical Field
The invention belongs to the field of detection, relates to a steel rail abrasion automatic detection device, and particularly relates to a steel rail abrasion automatic detection device which is based on a word line laser image processing and a microprocessor and can effectively detect the surface abrasion depth and width of a steel rail head.
Background
The railway is a main artery for transportation, and compared with other transportation modes, the heavy haul railway transportation is rapidly developed all over the world due to the characteristics of large transportation volume and low cost. In rail equipment, rails are the most important components to directly bear the load of a train and guide the running of wheels. Whether the technical state of the steel rail is intact directly influences whether the train can run safely, stably and uninterruptedly at the specified speed. Railway locomotives transmit driving and braking forces through friction between wheel and rail, which can cause rail wear. Along with the high-speed, heavy-load and high-density operation of a locomotive, the abrasion of a steel rail is increased rapidly, and particularly the abrasion of the inner side surface of a small-radius curve outer rail is serious.
The detection technology of the rail abrasion goes through the processes of simple and visual detection of ruler tools, detection of a digitizer and the like. At present, in China, main methods such as contact clamp measurement, eddy current detection and optical triangulation are available in the aspect of rail abrasion detection, detection results often depend on attitudes of detection workers and experience of instrument use, and the methods have the problems of low detection efficiency, low detection precision and the like, and can not meet the development requirements of high speed at present. Although a detection device for detecting the abrasion of the surface of the rail by using laser is available at present, accurate step-by-step detection cannot be realized, and other detection methods only stay at the theoretical research level.
Disclosure of Invention
In order to obtain a detection device capable of executing a rail abrasion detection method, the invention provides a rail abrasion detection device to execute the rail abrasion method, which is technically characterized in that: the device comprises a word line laser, a CCD image sensor and a microprocessor, wherein a plane where a word line light beam emitted by the word line laser is located and the surface of a steel rail to be detected form an angle of 60 degrees, the CCD image sensor is located right above the plane where the word line light beam is located, and the input end of the microprocessor receives steel rail image information collected by the CCD image sensor and executes a steel rail abrasion detection method.
Has the advantages that: the invention provides a steel rail abrasion detection device, and laser and the surface of a guide rail are irradiated at a certain angle, so that two sections of laser images formed on the surface are not on the same straight line in the direction vertical to the guide rail when the steel rail is abraded; the 60 degree is for two-segment images to form a special triangle on the wearing section, namely a 60 degree right triangle, so that the wearing width and depth can be conveniently calculated.
Drawings
Fig. 1 is a block diagram showing the structure of an automatic rail wear detection device according to embodiment 2;
FIG. 2 is a schematic view of a non-abrasive laser image;
FIG. 3 is a schematic view of a laser image with wear;
FIG. 4 is a graph showing the luminance curve and the diameter of the disk;
FIG. 5 is a labeled diagram of data points and feature quantities.
Detailed Description
Example 1: a rail abrasion automatic detection method comprises the steps of collecting a laser image of a rail, comparing the laser image with a complete rail laser light band image to judge whether the detected rail is abraded or not, judging whether the rail is abraded or not, carrying out laser image processing on the collected laser image, wherein the laser image processing comprises image preprocessing and image edge extraction, and after the laser image processing, selecting and extracting laser image characteristic quantities related to rail abrasion quantity to calculate rail abrasion depth and width. Wherein: the extracted laser image characteristic quantity related to the rail abrasion quantity is more than one of the following characteristic quantities:
1) length l of two linear portions of laser imageAAnd lB；
2) The width difference e of the two linear laser images;
3) longitudinal position difference z of the two linear laser images;
4) length l of transition section between two linear laser imagesC；
5) The inclination angle theta of a transition section between two sections of linear laser images;
the wear width and the wear depth are collectively called as the characteristic quantity of the rail wear, and one or more laser image characteristic quantities are selected and used for calculating the depth and the width of the rail wear;
when the depth and the width of the rail abrasion are calculated, not all the laser image characteristic quantities are selected for calculation, in order to optimize the calculation process, a combination of the laser image characteristic quantities is selected as basic calculation data for calculating the depth and the width of the rail abrasion, the combination is selected by firstly determining the laser image characteristic quantities related to the abrasion characteristics, selecting a preferred characteristic quantity from the laser image characteristic quantities, calculating correlation coefficients of the other laser image characteristic quantities and the preferred characteristic quantity, and calculating an average value of the correlation coefficients, wherein the average value is a threshold β for characteristic quantity selection, if an absolute value | rTij | β of the correlation coefficients between two laser image characteristics is equal to or more than a set of characteristic samples collected at fixed points on a rail, the set comprises the laser image characteristic quantities of the reaction of N fixed points, only one of the correlation coefficients is selected as the laser image characteristic quantity for rail abrasion judgment, namely, when the combination of the laser image characteristic quantities for judgment is selected, M is a characteristic sample set collected at the fixed points on the rail with the abrasion characteristics, the abrasion characteristics are selected as a measurement, and only one of the laser image characteristic quantity is selected as an effective parameter for the depth judgment of the rail abrasion characteristics of the laser image combination of the abrasion characteristics.
As an embodiment, the specific method for representing the correlation between the features by using the correlation coefficient as a measurement parameter is as follows: two different sets of laser image characteristics are used: t isi＝{tikK is 1,2, …, n and Tj＝{tjkAnd k is 1,2, …, n, where k denotes the kth test point and n test points are total, the correlation coefficient of the two groups of laser image features is defined as follows:
in the formula (I), the compound is shown in the specification,
and
are respectively two groups of characteristics TiAnd TjAverage value of (d):
and
correlation coefficient rTijReflects two groups of characteristics TiAnd TjDegree of correlation of (2), rTijWhen the value of (A) is negative, the negative correlation of the two characteristics is shown; rTijWhen the value of (a) is positive, the positive correlation of the two characteristics is shown; when rTijWhen 0, there is no correlation between the two wear characteristics, when rTijThe closer to 1 the absolute value of (A) is, the higher the correlation between the two laser image features and the greater the redundancy that occurs, and the more each laser is used in the laser image feature set that reflects rail wearThe correlation coefficient between the light image characteristic quantities is set to a threshold value β if the absolute value of the correlation coefficient | rT between two laser image characteristicsijAnd | not less than β, wherein the two laser image characteristics are related redundant characteristics, and only one of the two laser image characteristics is selected as the laser image characteristic quantity for judging the rail wear.
In another embodiment, the method for determining the threshold β is to select a single laser image feature quantity as a preferred feature, and determine the possibility that the remaining laser image feature quantities are redundant features by, after determining the preferred feature, obtaining a correlation coefficient between the preferred feature quantity and the laser image feature set in the rail wear width and depth correlation, setting the average value of the correlation coefficient data as the threshold β, and determining the threshold β by:
wherein: where c is the number of feature quantities, l is the number of preferred feature quantities, and j is the number of candidate feature quantities.
Thus, in the above embodiment, the correlation coefficient between the features is obtained to obtain the possibility of redundancy between the features, the average value of the set of correlation coefficient data is set as the threshold β, and the threshold is used as the basis for judging whether the features are redundant, so that when judging that two features are redundant, only one of the redundant features is selected as the laser image feature for calculating the depth and width of wear, so as to optimize the calculation process, thereby obtaining the minimum feature combination.
As an example, a method of calculating the wear depth and width is specifically disclosed: length l of two linear portions of laser imageAAs a preferred feature for abrasion width detection, the longitudinal position difference z of the two linear laser images is used as a preferred feature for abrasion depth detection; calculating the length l of two linear parts of the laser image according to the correlation coefficient of the two groups of laser image characteristicsAAnd lBThe longitudinal position difference z of the two linear laser images is used as the characteristic quantity of the abrasion width detection,the wear width calculation formula is:
wherein l is the width of the unworn steel rail;
the abrasion depth calculation formula is as follows:
V＝z·tan 60°
as an embodiment, the image preprocessing comprises the steps of:
firstly, graying an image, drawing a histogram of a grayscale image, and finding out a grayscale concentration range;
then, the gray level of the gray level image is enhanced by using the following formula, so that the image is clearer;
wherein: a. b are respectively the left and right boundary points of the gray value centralized distribution in the gray image histogram, and x and y respectively represent the gray values before and after the gray enhancement.
As an embodiment, the method for image edge extraction includes the following steps:
taking a median filtering brightness curve in which pixel points are distributed along the horizontal direction, respectively taking continuous points with the maximum brightness gradient change at two sides of the maximum peak value of the curve, and taking midpoints p and q of the two groups of continuous points, and taking the distance between p and q as the diameter of a detection template;
setting the brightness of the image as f (i, j), and taking a circle s (c, r) in the image field as a detection template, wherein c is the center of the circle and the coordinate is (i, j)c,jc) R is a radius;
defining a set of pixel points in s (c, r), and recording the brightness sum of the pixel points in the circle s as:
the circle center of the detection template is moved in a small range in the horizontal directionCalculating the brightness sum of each pixel in each position detection template, wherein the brightness sum of the maximum template circle center position in the range is a pixel-level ridge edge point of the bright strip, fitting a straight line by using a least square method, wherein the straight line is a central line of a laser image of a word line, and the small range is an image interval with the circle center as a central point and 2 times of radius respectively, so as to obtain the length l of two straight line parts of the laser imageAAnd lBLength l of transition section between two linear laser imagesC。
The laser image of the steel rail collected in the embodiment is compared with the complete image to judge whether the steel rail in the collected image is worn, when the steel rail is judged to be worn, the characteristic quantity is further selected to be used for calculating to obtain the wear, the wear is judged in a fixed mode, then the thought of quantitatively calculating the depth and the width of the wear is calculated, and in the calculating process, the characteristic quantity is selected to optimize the calculating process of the depth and the width of the wear.
Example 2: as a supplement to the solution of embodiment 1, or as a separate embodiment: the abrasion mainly occurs on the head of the steel rail, the abrasion comprises top surface abrasion and side surface abrasion, and the two numerical values must be detected simultaneously during detection to comprehensively judge the abrasion degree of the steel rail. In the embodiment, a high-intensity narrow-beam linear laser beam is utilized, an angle of 60 degrees is formed between the plane where the laser and the linear laser beam are located and the surface of the measured steel rail, and the high-resolution area array CCD image sensor is located right above the laser image to shoot the laser image. The light beam image shows bending on the surface of the worn steel rail, and the width and the depth of the rail wear are determined by the position and the bending degree of the bending point.
The rail wearing and tearing automatic checkout device includes: the device comprises a linear laser, a CCD image sensor, a microprocessor, an execution unit, a display and acousto-optic alarm unit and an interface unit. The CCD image sensor collects laser images, the obtained image information is transmitted to the microprocessor for analysis and processing, the edge and the center of the image are extracted, a straight line is fitted, a complete steel rail laser light band image outline is formed, the image information is converted into steel rail outline parameters, the characteristic quantity of the steel rail outline is stored and compared with the complete steel rail parameters, and whether the steel rail has abrasion or not is judged. Continuing to detect the next point when no abrasion is generated; and (4) determining abrasion loss, including the depth and width of abrasion loss. The execution unit receives a control signal of the microprocessor, controls the advancing direction and the speed of the detection device, and adjusts the orientation of the CCD image sensor, the output end of the microprocessor is respectively connected with the LCD display and the acousto-optic alarm system, the LCD display is used for displaying the current position and the abrasion degree of the steel rail, and the acousto-optic alarm system is used for prompting that the current position of the steel rail is abraded and needs to be repaired. The interface unit is used for exchanging information with an upper computer, and the upper computer can further finely process the image of the abrasion position and determine the accurate abrasion amount.
The image preprocessing is a preprocessing stage of laser image edge extraction, firstly graying an image, drawing a histogram of the grayscale image, finding out a grayscale concentration range, and performing grayscale enhancement on the grayscale image by using a formula (1) (wherein a and b are respectively left and right boundary points of the grayscale value concentration distribution in the histogram of the grayscale image, and x and y respectively represent grayscale values before and after grayscale enhancement) so as to make the image clearer.
The edge detection of the laser image of the first line adopts a roof ridge edge detection method. The edge detection method based on the brightness of a single pixel point has poor noise resistance, and in order to reduce the interference of image noise, the brightness sum of each pixel point in a certain area is used as a ridge-shaped edge judgment basis. Because the circle has isotropy and is not influenced by the ridge edge direction, the invention adopts a ridge edge detection method by a disc method. And moving a disc detection template with a proper size in a certain range at two sides of the laser image of the same line, wherein when the brightness and gradient change of each pixel point in the template meet certain requirements, the central point of the template is a ridge-shaped edge point.
Any one median filtering brightness curve with pixel points distributed along the horizontal direction is taken, continuous points with the maximum brightness gradient change are taken out from two sides of the maximum peak value of the curve respectively, the middle points p and q of the two groups of continuous points are taken, and the distance between the points p and q is taken as the diameter of the detection template, as shown in fig. 3.
Setting the brightness of the image as f (i, j), and taking a circle s (c, r) in the image field as a detection template, wherein c is the center of the circle and the coordinate is (i, j)c,jc) And r is a radius. Defining a set of pixel points within s (c, r):
and recording the brightness sum of the pixel points in the circle s as:
and moving the circle center of the detection template in a small range in the horizontal direction, and calculating the brightness sum of each pixel in each position detection template, wherein the position of the circle center of the template with the maximum brightness sum in the range is a pixel-level ridge edge point of the bright strip. And fitting a straight line by using a least square method, wherein the straight line is the central line of the laser image of the word line. The detected edge points and the fitted straight line are shown in fig. 4.
And further extracting characteristic quantities of the laser image related to the rail abrasion loss, including a selection method of the characteristic quantities and determination of a threshold value.
The width and the depth of the rail abrasion, which are mainly related by the invention, can be determined through the bending degree of the laser image, and the following characteristic quantities can be used for selection:
1) length l of two linear portions of laser imageAAnd lB；
2) The width difference e of the two linear laser images;
3) longitudinal position difference z of the two linear laser images;
4) length l of transition section between two linear laser imagesC；
5) And the inclination angle theta of the transition section between the two linear laser images.
One or more characteristic quantities can be selected for judging the depth and the width of the wear of the steel rail, and when the combination of the characteristic quantities for judging is selected, different types of characteristics are required to have obvious difference, so that redundant characteristic interference judgment is avoided. Assuming that M is spot mining on a worn railA set of feature samples, the set comprising wear features for n fixed points. And selecting the correlation coefficient as a measurement parameter, wherein the parameter can reflect the similarity between the characteristics and is used for searching the combination of the minimum characteristic quantity which can effectively judge the wear width and the depth of the steel rail. Two different sets of wear characteristics were set as: t isi＝{tikK is 1,2, …, n and Tj＝{tjkAnd k is 1,2, …, n, where k denotes the kth test point and n test points are total, the correlation coefficients of the two sets of features are defined as follows:
in the formula (I), the compound is shown in the specification,
and
are respectively two groups of characteristics TiAnd TjAverage value of (d):
and
correlation coefficient rTijReflects two groups of characteristics TiAnd TjDegree of correlation of (2), rTijWhen the value of (A) is negative, the negative correlation of the two characteristics is shown; and if the value is positive, the positive correlation of the two characteristics is shown. When rTijWhen 0, there is no correlation between the two features. When rT is thenijThe closer the absolute value of (a) is to 1, the higher the degree of correlation between the two features is, and the greater the redundancy that can be generated in this case.
In the feature set of rail wear, a threshold value β is set using the correlation coefficient between the respective feature quantities, and if the absolute value | rT of the correlation coefficient between two of the feature quantities is greater than the threshold value | rTij| is equal to or more than β, the two characteristics are related redundant characteristics, and only one of the characteristics can be selected as the characteristic quantity for judging the rail wear。
After the preferred features are determined, a correlation coefficient between the preferred feature quantity and the rail wear width in a related feature set is obtained through calculation, and the mean value of the group of data is set as a threshold β, wherein the average value is shown in formula (4):
and after the characteristic quantity is determined, calculating the steel rail abrasion quantity of the detection position, storing and displaying, and starting an audible and visual alarm device when the abrasion exceeds the limit.
Under the off-line condition, the interface unit is used for exchanging information with the upper computer, and the upper computer can further finely process the image of the abrasion position to determine the accurate abrasion amount. By adopting the technical scheme, the automatic steel rail wear detection device provided by the embodiment has the beneficial effects that the image processing method is adopted, the control of the microprocessor is separated from the control of a PC, and the device can automatically run under the setting of an operator. The device has certain perfection and effectiveness, is convenient for detection personnel to use, and has the advantages of simple operation, accurate detection result and low production and manufacturing cost.
Example 3: a steel rail abrasion detection device comprises a word line laser, a CCD image sensor and a microprocessor, wherein a plane where a word line light beam emitted by the word line laser is located and the surface of a steel rail to be detected form an angle of 60 degrees, the CCD image sensor is located right above the plane where the word line light beam is located, the input end of the microprocessor receives steel rail image information collected by the CCD image sensor and executes a steel rail abrasion detection method, and the detection method can be the detection method in embodiment 1. The laser irradiates the surface of the guide rail at a certain angle, and two sections of laser images formed on the surface are not on the same straight line in the direction vertical to the guide rail in order to ensure that the steel rail is worn; the 60 degree is for two-segment images to form a special triangle on the wearing section, namely a 60 degree right triangle, so that the wearing width and depth can be conveniently calculated.
As an embodiment, the detection device further comprises an LCD display, an acousto-optic alarm system and an interface unit, wherein the output end of the microprocessor is respectively connected with the LCD display, the acousto-optic alarm system and the interface unit, and the interface unit is connected with the upper computer to exchange information between the microprocessor and the upper computer.
As an embodiment, the detection device further comprises an execution unit, the execution unit receives a control signal of the microprocessor, the execution unit comprises a first servo motor and a second servo motor, the microprocessor is connected with the first servo motor to control the advancing direction and the advancing speed of the steel rail abrasion detection system, the microprocessor is connected with the second servo motor, the second servo motor is connected with the CCD image sensor, and the second servo motor is used to control the orientation of the CCD image sensor. The processor sends out a control signal to control a servo motor of the execution unit to fix time according to a fixed speed form, a next detection point is determined, the angle of the CCD image sensor is finely adjusted through the motor, and the laser light and the surface of the steel rail form an angle of 60 degrees.
As an example: the microprocessor includes:
the abrasion judging module is used for acquiring a laser image of the steel rail and comparing the laser image with a complete steel rail laser light band image to judge whether the steel rail is abraded or not;
and the abrasion calculation module is used for judging the abrasion of the steel rail, selecting and extracting the laser image characteristic quantity related to the abrasion quantity of the steel rail so as to calculate and obtain the abrasion depth and/or width of the steel rail.
As an embodiment, the detection apparatus further includes:
the characteristic quantity selection module is used for selecting the laser image characteristic quantity related to the rail abrasion quantity to be more than one of the following characteristic quantities:
1) two segments of laser imageLength l of straight line partAAnd lB；
2) The width difference e of the two linear laser images;
3) longitudinal position difference z of the two linear laser images;
4) length l of transition section between two linear laser imagesC；
5) The inclination angle theta of a transition section between two sections of linear laser images;
the process of selecting the feature quantity by the feature quantity selection module is as follows: the characteristic quantity of the rail abrasion comprises an abrasion width and an abrasion depth, one or more laser image characteristic quantities are selected and used for calculating the depth and/or the width of the rail abrasion, when the combination of the laser image characteristic quantities used for judging is selected, if M is a characteristic sample set collected at fixed points on the rail with the abrasion, the set comprises laser image characteristic quantities of reaction abrasion of N fixed points, a correlation coefficient is selected as a measurement parameter, two laser image characteristics are related redundant characteristics, only one laser image characteristic quantity is selected as the laser image characteristic quantity used for judging the rail abrasion, so that the combination of the minimum laser image characteristic quantities for effectively judging the rail abrasion width and the depth is obtained, and a certain characteristic quantity of the rail abrasion is calculated by the characteristic quantity combination.
Selecting a preferred characteristic quantity from the laser image characteristic quantities, calculating the correlation coefficient of each of the other laser image characteristic quantities and the preferred characteristic quantity, and solving an average value of the correlation coefficients, wherein the average value is a threshold β selected by the laser image characteristic quantities, and if the absolute value | rTij | of the correlation coefficient between certain two laser image characteristics is more than or equal to β, the two laser image characteristics are related redundant characteristics, and only one of the two laser image characteristics is selected as the laser image characteristic quantity for judging the rail abrasion.
The method for calculating the correlation coefficient comprises the following steps:
two different sets of laser image characteristics are respectively: t isi＝{tikK is 1,2, …, n and Tj＝{tjkAnd k is 1,2, …, n, where k denotes the k-th test point and n test points are total, the correlation of the two groups of laser image featuresThe coefficients are defined as follows:
in the formula (I), the compound is shown in the specification,and
are respectively two groups of characteristics TiAnd TjAverage value of (d):and
correlation coefficient rTijReflects two groups of characteristics TiAnd TjDegree of correlation of (2), rTijWhen the value of (A) is negative, the negative correlation of the two characteristics is shown; rTijWhen the value of (a) is positive, the positive correlation of the two characteristics is shown; when rTijWhen 0, there is no correlation between the two wear characteristics, when rTijWhen the absolute value of (1) is closer to 1, the correlation degree of the two laser image characteristics is higher, and the generated redundancy is higher;
the method of calculating threshold β is:
wherein: where c is the number of feature quantities, l is the number of preferred feature quantities, and j is the number of candidate feature quantities.
Length l of two linear portions of laser imageAAs a preferred feature for abrasion width detection, the longitudinal position difference z of the two linear laser images is used as a preferred feature for abrasion depth detection; calculating the length l of two linear parts of the laser image according to the correlation coefficient of the two groups of laser image characteristicsAAnd lBTwo-stage linear laser map as characteristic quantity for wear width detectionThe longitudinal position difference z of the image is used as a characteristic quantity of abrasion depth detection, and an abrasion width calculation formula is as follows:
wherein l is the width of the unworn steel rail;
the abrasion depth calculation formula is as follows:
V＝z·tan 60°。
the method comprises the steps of collecting a laser image of a steel rail, and before extracting laser image characteristic quantity related to the abrasion quantity of the steel rail, carrying out laser image processing, wherein the laser image processing comprises image preprocessing and image edge extraction;
the image preprocessing comprises the following steps:
firstly, graying an image, drawing a histogram of a grayscale image, and finding out a grayscale concentration range;
then, the gray level of the gray level image is enhanced by using the following formula, so that the image is clearer;
wherein: a. b are respectively left and right boundary points of the gray value centralized distribution in the gray image histogram, and x and y respectively represent the gray values before and after gray enhancement;
the method for extracting the image edge comprises the following steps:
taking a median filtering brightness curve in which pixel points are distributed along the horizontal direction, respectively taking continuous points with the maximum brightness gradient change at two sides of the maximum peak value of the curve, and taking midpoints p and q of the two groups of continuous points, and taking the distance between p and q as the diameter of a detection template;
setting the brightness of the image as f (i, j), and taking a circle s (c, r) in the image field as a detection template, wherein c is the center of the circle and the coordinate is (i, j)c,jc) R is a radius;
defining a set of pixel points in s (c, r), and recording the brightness sum of the pixel points in the circle s as:
moving the circle center of the detection template in a small range in the horizontal direction, calculating the brightness sum of each pixel in each position detection template, wherein the brightness sum in the range is the maximum template circle center position, namely a pixel-level ridge edge point of the bright strip, fitting a straight line by using a least square method, wherein the straight line is a central line of a laser image of a word line, and the small range is an image interval with the circle center as the central point and 2 times of radius respectively, so as to obtain the length l of two straight line parts of the laser imageAAnd lBLength l of transition section between two linear laser imagesC。
The above description is only for the purpose of creating a preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can substitute or change the technical solution and the inventive concept of the present invention within the technical scope of the present invention.
Claims (1)
1. A method for judging two laser image features as related redundant features includes selecting a preferred feature from laser image features, calculating correlation coefficient between the other laser image features and the preferred feature, calculating average value as threshold β for selecting laser image features, judging whether two laser image features are related redundant features if absolute value of correlation coefficient between two laser image features is | rTij | ≧ β, and selecting only one of the two laser image features as judged laser image feature;
two different sets of laser image characteristics are used: t isi＝{tikK is 1,2, …, n and Tj＝{tjkAnd k is 1,2, …, n, where k denotes the kth test point and n test points are total, the correlation coefficient of the two groups of laser image features is defined as follows:
in the formula (I), the compound is shown in the specification,
and
are respectively two groups of characteristics TiAnd TjAverage value of (d):
andcorrelation coefficient rTijReflects two groups of characteristics TiAnd TjDegree of correlation of (2), rTijWhen the value of (A) is negative, the negative correlation of the two characteristics is shown; rTijWhen the value of (a) is positive, the positive correlation of the two characteristics is shown; when rTijWhen 0, there is no correlation between the two wear characteristics, when rTijThe closer to 1 the absolute value of (a) is, the higher the correlation between the two laser image features becomes, and the greater the redundancy is.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711422597.9A CN108189859B (en) | 2016-08-30 | 2016-08-30 | Method for judging two laser image characteristics as related redundant characteristics |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610765945.1A CN106274979B (en) | 2016-08-30 | 2016-08-30 | Rail wear automatic detection device |
| CN201711422597.9A CN108189859B (en) | 2016-08-30 | 2016-08-30 | Method for judging two laser image characteristics as related redundant characteristics |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610765945.1A Division CN106274979B (en) | 2016-08-30 | 2016-08-30 | Rail wear automatic detection device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108189859A CN108189859A (en) | 2018-06-22 |
| CN108189859B true CN108189859B (en) | 2020-02-14 |
Family
ID=57675472
Family Applications (4)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711419564.9A Expired - Fee Related CN108128323B (en) | 2016-08-30 | 2016-08-30 | The selection method of laser image characteristic quantity relevant to rail wear amount |
| CN201610765945.1A Expired - Fee Related CN106274979B (en) | 2016-08-30 | 2016-08-30 | Rail wear automatic detection device |
| CN201711421370.2A Active CN108177660B (en) | 2016-08-30 | 2016-08-30 | Steel rail abrasion detection method with laser image processing step |
| CN201711422597.9A Expired - Fee Related CN108189859B (en) | 2016-08-30 | 2016-08-30 | Method for judging two laser image characteristics as related redundant characteristics |
Family Applications Before (3)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711419564.9A Expired - Fee Related CN108128323B (en) | 2016-08-30 | 2016-08-30 | The selection method of laser image characteristic quantity relevant to rail wear amount |
| CN201610765945.1A Expired - Fee Related CN106274979B (en) | 2016-08-30 | 2016-08-30 | Rail wear automatic detection device |
| CN201711421370.2A Active CN108177660B (en) | 2016-08-30 | 2016-08-30 | Steel rail abrasion detection method with laser image processing step |
Country Status (1)
| Country | Link |
|---|---|
| CN (4) | CN108128323B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108662983B (en) * | 2016-08-30 | 2020-05-01 | 大连民族大学 | Method for detecting and calculating correlation coefficient of steel rail abrasion |
| CN107194921B (en) * | 2017-05-19 | 2020-04-07 | 华夏高铁技术有限公司 | Automatic identification method and device for rail surface light band of steel rail |
| CN111457835A (en) * | 2020-03-30 | 2020-07-28 | 合肥康普曼数字技术有限公司 | Threaded hole tooth socket detection method for battery tray |
| CN113172551B (en) * | 2020-05-29 | 2022-10-14 | 浙江大学 | Quantitative measurement method for surface machining quality of steel rail |
| CN111830049B (en) * | 2020-07-24 | 2023-09-22 | 深圳市道通科技股份有限公司 | Brake disc wear area identification method and wear identification system |
| CN113077437B (en) * | 2021-03-31 | 2023-07-25 | 上海晨兴希姆通电子科技有限公司 | Workpiece quality detection method and system |
| CN116047020B (en) * | 2023-04-03 | 2023-07-25 | 杭州电子科技大学 | Online repair performance testing device and testing method for steel rail surface damage |
| CN117029696B (en) * | 2023-10-08 | 2024-01-02 | 天津津铁供电有限公司 | Abrasion detection method and detection equipment for rigid suspension contact net |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101144714A (en) * | 2007-10-26 | 2008-03-19 | 北京航空航天大学 | Steel rail wearing integrative parameter vehicle-mounted dynamic measuring device and method |
| CN101576375A (en) * | 2009-05-21 | 2009-11-11 | 北京航空航天大学 | Fast processing method of laser vision image of steel rail wear |
| CN102749061A (en) * | 2012-07-26 | 2012-10-24 | 上海工程技术大学 | Steel rail abrasion measuring method based on dynamic template |
Family Cites Families (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4198164A (en) * | 1976-10-07 | 1980-04-15 | Ensco, Inc. | Proximity sensor and method and apparatus for continuously measuring rail gauge |
| JPH0611331A (en) * | 1991-12-26 | 1994-01-21 | Tokimec Inc | Instrument and method for measuring undulating wear of rail |
| JP4302288B2 (en) * | 2000-04-19 | 2009-07-22 | 東京計器株式会社 | Portable rail cross-section measuring device |
| ITVE20000023A1 (en) * | 2000-05-12 | 2001-11-12 | Tecnogamma S A S Di Zanin E & | LASER EQUIPMENT FOR THE CONTROL OF THE RAILWAYS OF A RAILWAY LINE. |
| AT5911U3 (en) * | 2002-10-29 | 2003-11-25 | Plasser Bahnbaumasch Franz | METHOD FOR CONTACT-FREE MEASUREMENT OF A CROSS-PROFILE OR DISTANCE FROM RAILS OF A TRACK |
| DE10326033B4 (en) * | 2003-06-10 | 2005-12-22 | Hema Electronic Gmbh | Method for adaptive error detection on an inhomogeneous surface |
| CN100462678C (en) * | 2005-11-22 | 2009-02-18 | 北京航空航天大学 | Steel rail near laser visual dynamic measuring device and method |
| CN101532827B (en) * | 2009-04-15 | 2010-12-01 | 北京航空航天大学 | Deviation correction method for measurement of rail wear based on laser vision |
| CN102175953A (en) * | 2011-02-28 | 2011-09-07 | 长沙理工大学 | Discharge online monitoring and fault positioning method for power transmission line insulator |
| CN202400107U (en) * | 2011-10-19 | 2012-08-29 | 北京鼎汉检测技术有限公司 | Detection device for dynamically detecting abrasion of lateral sides of steel railway rails |
| CN102565068A (en) * | 2011-12-12 | 2012-07-11 | 关持循 | Steel rail damage automatic detection device |
| CN103171589B (en) * | 2011-12-26 | 2016-08-10 | 同济大学 | A kind of Rail exterior feature detection method based on wearing valve |
| CN104501768B (en) * | 2014-08-14 | 2017-05-31 | 武汉武大卓越科技有限责任公司 | Orbit rigidity measuring method based on machine vision |
| CN104296682A (en) * | 2014-10-28 | 2015-01-21 | 电子科技大学 | Contour registration method used in train guide rail contour measurement based on machine vision |
| CN104442917A (en) * | 2014-12-31 | 2015-03-25 | 北京二七轨道交通装备有限责任公司 | Track detection method |
| CN104899888B (en) * | 2015-06-18 | 2017-10-24 | 大连理工大学 | A kind of image sub-pixel edge detection method based on Legendre squares |
-
2016
- 2016-08-30 CN CN201711419564.9A patent/CN108128323B/en not_active Expired - Fee Related
- 2016-08-30 CN CN201610765945.1A patent/CN106274979B/en not_active Expired - Fee Related
- 2016-08-30 CN CN201711421370.2A patent/CN108177660B/en active Active
- 2016-08-30 CN CN201711422597.9A patent/CN108189859B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101144714A (en) * | 2007-10-26 | 2008-03-19 | 北京航空航天大学 | Steel rail wearing integrative parameter vehicle-mounted dynamic measuring device and method |
| CN101576375A (en) * | 2009-05-21 | 2009-11-11 | 北京航空航天大学 | Fast processing method of laser vision image of steel rail wear |
| CN102749061A (en) * | 2012-07-26 | 2012-10-24 | 上海工程技术大学 | Steel rail abrasion measuring method based on dynamic template |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108189859A (en) | 2018-06-22 |
| CN108177660A (en) | 2018-06-19 |
| CN108177660B (en) | 2020-07-14 |
| CN106274979A (en) | 2017-01-04 |
| CN108128323A (en) | 2018-06-08 |
| CN106274979B (en) | 2018-06-22 |
| CN108128323B (en) | 2019-09-17 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108189859B (en) | Method for judging two laser image characteristics as related redundant characteristics | |
| CN108731599B (en) | Steel rail abrasion depth calculation method | |
| CN109580657B (en) | Crack detection method in bridge quality detection | |
| CN103217111B (en) | A kind of non-contact contact line geometric parameter detection method | |
| CN109969736A (en) | A kind of large size carrier strip deviation fault intelligent detecting method | |
| JP2017187418A (en) | Wear inspection apparatus and wear inspection method | |
| CN111982916A (en) | Welding seam surface defect detection method and system based on machine vision | |
| CN110567680B (en) | Track fastener looseness detection method based on angle comparison | |
| CN111311560A (en) | Method and device for detecting state of steel rail fastener | |
| CN113448333A (en) | Bottom routing inspection positioning method and device based on sensor combination and electronic equipment | |
| CN206019585U (en) | Rail Abrasion Detection System system | |
| CN110490342B (en) | Contact net static geometrical parameter detection method based on Faster R-CNN | |
| CN110186934B (en) | Axle box rubber pad crack detection method and detection device | |
| CN116309598B (en) | Intelligent grading method for steel continuous casting billet macrostructure cracks based on machine vision | |
| CN112945976B (en) | Method and device for detecting contact fatigue crack of steel rail | |
| CN116681912A (en) | Rail gauge detection method and device for railway turnout | |
| CN113222907B (en) | Detection robot based on curved rail | |
| CN111127459B (en) | Real-time image processing system for detecting ruts on road surface | |
| JP6389783B2 (en) | Crossover relative position management apparatus and method | |
| JP2008281461A (en) | Trolley wire wear measuring apparatus by image processing | |
| CN115908400A (en) | Intelligent detection and positioning method for track irregularity and diseases | |
| CN117876368B (en) | Method and system for detecting abrasion and crack of carbon slide plate of contact rail collector shoe | |
| CN112651988B (en) | Finger-shaped image segmentation, finger-shaped plate staggering and abnormal fastener detection method based on double-pointer positioning | |
| CN115035142A (en) | Automatic subway tunnel track line identification method based on point cloud | |
| CN112406948A (en) | Measurement system for full profile wearing and tearing disease to rail railhead |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200214 |