CN106840033A - A kind of profile of steel rail detection means and method based on image procossing - Google Patents

A kind of profile of steel rail detection means and method based on image procossing Download PDF

Info

Publication number
CN106840033A
CN106840033A CN201710147303.XA CN201710147303A CN106840033A CN 106840033 A CN106840033 A CN 106840033A CN 201710147303 A CN201710147303 A CN 201710147303A CN 106840033 A CN106840033 A CN 106840033A
Authority
CN
China
Prior art keywords
sensor
profile
data
rotation
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.)
Granted
Application number
CN201710147303.XA
Other languages
Chinese (zh)
Other versions
CN106840033B (en
Inventor
李存荣
谢雯
张琪
崔耕
王博文
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuhan University of Technology WUT
Original Assignee
Wuhan University of Technology WUT
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Wuhan University of Technology WUT filed Critical Wuhan University of Technology WUT
Priority to CN201710147303.XA priority Critical patent/CN106840033B/en
Publication of CN106840033A publication Critical patent/CN106840033A/en
Application granted granted Critical
Publication of CN106840033B publication Critical patent/CN106840033B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01BMEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
    • G01B11/00Measuring arrangements characterised by the use of optical means
    • G01B11/24Measuring arrangements characterised by the use of optical means for measuring contours or curvatures
    • G01B11/245Measuring arrangements characterised by the use of optical means for measuring contours or curvatures using a plurality of fixed, simultaneously operating transducers

Abstract

The present invention relates to a kind of profile of steel rail detection means and method based on image procossing.The device includes four sensors, sensor stand (5), base (6), orbit transports roller (7) and rail (8), four sensors are separately mounted at four interior angles of sensor stand, the Z axis of each sensor coordinate system intersect at the center of the sensor stand being fixed on base, orbit transports roller is fixed with base, for supporting and conveying rail.The method is that the image for collecting four sensors carries out coordinate transform, unify in same world coordinate system, using one of image as benchmark, characteristic point in four images of search, according to characteristic point by its excess-three image to Reference Transforming, rail profile profile is obtained, the physical dimension of profiled outline is calculated and is compared judgement with iron mark.The method can significantly improve production efficiency, certainty of measurement is improved.

Description

A kind of profile of steel rail detection means and method based on image procossing
Technical field
The present invention relates to railway rail detection technique field, and in particular to a kind of profile of steel rail parameter based on image procossing Measurement apparatus and method.
Background technology
With the continuous lifting of train running speed and density, stationarity, comfortableness to train operation, security requirement Continuous improvement, influence factor present in the solder technology to rail is it is also proposed that carry out requirements at the higher level.Seamless welding technology into It is main flow, in weld job, the physical dimension of rail profile profile is the key factor of the glacing flatness for influenceing rail.
Rail welding plant is in rail welding operation is carried out, if the size of weld both ends of the surface is inconsistent will to cause rail quality Decline, increase polishing workload, and influence traffic safety.
The matching precision that current profile of steel rail measurement method of parameters has profile of steel rail is not high, have impact on certainty of measurement, The present invention has carried out the computing of multiple characteristic point during being demarcated, and extracts relative accuracy characteristic point higher, and Using linearity rectangle calibrated bolck high, the accuracy and precision of demarcation are improve.
The content of the invention
The technical problem to be solved in the present invention is:In view of the above, there is provided a kind of profile of steel rail based on image procossing Detection means and method, to improve the precision of Image Mosaic, so as to improve the accuracy of detection of profile of steel rail size.
The present invention solves its technical problem and uses following technical scheme:
The profile of steel rail detection means based on image procossing that the present invention is provided, it is characterized in that including four sensors, passing Sensor bracket, base, orbit transports roller and rail, wherein:Four sensors are laser profile sensor, and they are separately mounted to At four interior angles of sensor stand, and the Z axis of each sensor coordinate system intersect at the center of sensor stand, sensor Support is fixed on the base of the detection means, while orbit transports roller is fixed with base, for supporting and conveying rail.
Four described sensors, using the linear laser distance measuring sensor of same model.
The profile of steel rail detection method based on image procossing that the present invention is provided, specifically:It is imaged using non-contact laser Principle, gathers rail profile outline data, by what is collected to each sensor using four three-dimensional laser profile sensors View data carries out coordinate transform, rotation, translation, carries out pattern conjunction, obtains rail profile profile, the section that will be calculated Profile physical dimension is compared judgement with iron mark.
It is first disconnected as rail with rectangular block before the measurement of profile of steel rail parameter is carried out using the data in the above method The calibrated bolck of facial contour detection carries out data scaling, calculates anglec of rotation R1, R2, R3, R4 and translational movement S1, S2, S3, S4, with With regard to that without being demarcated again, only need to be placed on rail on measurement station, directly using calibrating parameters by each measurement afterwards:Rotation Angle and translational movement obtain complete rail profile profile, to improve profile of steel rail detection efficiency.
In the above method, it can be demarcated using the preferable cuboid calibrated bolck of the linearity, specifically:By four Sensor is symmetrically mounted on axis in 45 ° of positions, and cuboid calibrated bolck centre-height is symmetrically installed with four sensors The foursquare center for being formed highly is consistent, and the image for then collecting four sensors carries out coordinate rotation and becomes Treatment is changed, it is unified in same world coordinate system, using one of image as benchmark, find out the feature in four images Point, translates its excess-three image to benchmark according to characteristic point, carries out pattern conjunction, obtains calibrated bolck profiled outline, is marked Fixed number evidence:The anglec of rotation and translational movement.
In the above method, during data scaling is carried out, may comprise steps of:
(1) it is placed on detection station using the calibrated bolck of intended size, opens laser sensor;
(2) data coordinate system by four laser sensor outputs carries out rotation transformation, is unified to the same world In coordinate system, calibrating parameters 1 are obtained, it is anglec of rotation R1, R2, R3, R4;
(3) postrotational data are carried out into coordinate translation according to the size of calibrated bolck, obtains calibrating parameters 2, it is data Translational movement S1, S2, S3, S4.
In the above method, described rotation transformation treatment comprises the following steps method:
Step 1, marks the starting point of gathered data, the top margin of the size criterion block according to calibrated bolck, base and side Side;
Step 2, linear fit is carried out by the top margin of acquisition or base:
In four laser sensors, the base of first sensor and 3rd sensor carries out linear fit, second sensor Top margin with the 4th sensor carries out linear fit, due to four X-direction precision of sensor be 0.15mm, take top margin or The data on the base at least 5mm of the end points away from side, the fitting of the work of the series data after 50 points is respectively given up directly to take away from end points and flex point Line;
Step 3:Obtain four slope of fitting a straight line respectively K1、K2、K3、K4, then obtain the inclination angle of fitting a straight line Degree θ=tan-1K, angle of inclination is designated as respectively:θ1, θ2, θ3, θ4
The relative position on base or top margin and side according to each image, the anglec of rotation is respectively R1=180 ° of-θ1, R2 =180 ° of-θ2, R3=-θ3, R4=-θ4, the anglec of rotation is substituted into coordinate rotation formula respectively:
|x′i=xoffset+xi×cosR-yi×sinR
y′i=yoffset+xi×sin R+yi×cos R
In formula:R is the anglec of rotation, xoffsetIt is the translational movement of X-direction, yoffsetIt is the translational movement of Y-direction, (xi, yi) it is biography The initial data point coordinates that sensor is collected, (x 'i, y 'i) it is coordinate (xi, yi) rotated with anglec of rotation R, and with vector (xoffset, yoffset) point coordinates after translation, before data translational movement is calculated, (x in formulaoffset, yoffset)=(0,0);
Step 4:The image after every group of data rotation is obtained, and same world coordinate system is arrived into the unification of postrotational data In.
In the above method, four coordinate systems can be carried out by translation treatment according to postrotational image, specifically by one The image of sensor keeps constant, and the image after its excess-three sensor rotation is translated to the image of first sensor, Spliced calibrated bolck profiled outline figure is obtained, so as to obtain calibrating parameters 2, i.e. data translational movement S1, S2, S3, S4;
The translation treatment is comprised the following steps:
Step 1:The flex point of second sensor is moved to and is overlapped with the X-coordinate of the flex point of first sensor, Z coordinate is The upward length of Z coordinate of the flex point of one sensor is that the point of calibrated bolck distance wide overlaps, and is sensed with translational movement translation second The view data of device;The flex point of 3rd sensor is moved to and is overlapped with the Z coordinate of the flex point of first sensor, X-coordinate is Length is the point coincidence of the distance of standard block length to the X-coordinate of the flex point of one sensor to the right, is sensed with translational movement translation the 3rd The view data of device;The flex point of the 4th sensor moves to the X-coordinate of the flex point of first sensor length is standard to the right The distance of block length, the upward length of Z coordinate of the flex point of first sensor is the point coincidence of calibrated bolck distance wide, with the translational movement Translate the view data of the 4th sensor;
Step 2:The fitting a straight line of the side that each sensor collects is sought with the method for linear fit in above-mentioned calibration process, The friendship of the top margin of the initial data that each sensor is collected or the fitting a straight line on base and the fitting a straight line of side is calculated respectively Point, and the point and the corresponding anglec of rotation are substituted into coordinate rotation formula, obtain postrotational flex point Zmax1, Zmax2, Zmax3, Zmax4
Step 3:The initial data that each sensor is collected carries out multi collect, gathers n times, takes out N group data By the flex point Z of step 2 gainedmax1, Zmax2, Zmax3, Zmax4, its average value is sought respectively, obtain Avgmax1, Avgmax2, Avgmax3, Avgmax4
Step 4:The translational movement of data is S after the rotation of first sensor1=0, data after the rotation of second sensor Translational movement is S2=Avgmax2(x,z)-Avgmax1(x, z+b), the translational movement of data is S after the rotation of 3rd sensor3= Avgmax3(x,z)-Avgmax1(x+a, z), the translational movement of data is S after the rotation of the 4th sensor4=Avgmax4(x,z)- Avgmax1(x+a,z+b)。
The above-mentioned method that the present invention is provided, it is described by the application in this rail profile profile measurement its profile size Profile size includes that rail is high, rail head is wide, the web of the rail is thick, flange of rail edge thickness, the flange of rail are wide, section degree of asymmetry.
It is to calculate former according to GB TB/T3276-2011, TB/T2344-2012 profile size regulation during this method application Reason, the rail profile profile removal after the completion of split is repeated to obtain the region of data, wide with cubic spline curve piecewise fitting Graphic data, calculates profile size on SPL, acquired results repeatability preferably, certainty of measurement and repeatable accuracy be above ± 0.03。
The present invention has the advantages that following main compared with prior art:
1. controllable using the accuracy of manufacture, make the good rectangle calibrated bolck of the simple linearity and demarcated as calibrating block, Simplify calibration process, be easy to implement automatic measurement;
2. the anglec of rotation is obtained using the method for high-precision sensor (linear laser distance measuring sensor) and linear fit and put down Shifting amount, improves the accuracy of measurement;
3. multiple coordinate system conversion need not be carried out, only with the profile of steel rail data that Cubic Spline Fitting is obtained, steel is detected The profile size of rail, reduces accumulated error, improves certainty of measurement;
4. demarcated using the preferable calibrated bolck of the linearity, can automatically be compensated installation and foozle.
Brief description of the drawings
Fig. 1 is the structural representation of profile of steel rail detection means of the present invention based on image procossing.
Fig. 2 is that the calibrated bolck of profile of steel rail detection method of the present invention based on image procossing demarcates preceding schematic diagram.
Fig. 3 is after the present invention is based on the calibrated data rotation of calibrated bolck of the profile of steel rail detection method of image procossing Design sketch.
Fig. 4 is the demarcation of the spliced calibrated bolck of calibrated bolck of profile of steel rail detection method of the present invention based on image procossing Result figure.
Fig. 5 is the rail profile profile measured result figure of profile of steel rail detection method of the present invention based on image procossing.
In figure:1. first sensor, 2. second sensor, 3. 3rd sensor, 4. the 4th sensor, 5. sensor branch Frame, 6. base, 7. orbit transports roller, 8. rail.
Specific embodiment
With reference to embodiment and accompanying drawing, the invention will be further described, but does not limit the present invention.
The present invention provides profile of steel rail detection detection means and method based on image procossing, using the device shown in Fig. 1 Realize, the device gathers outline data using the principle of triangulation of laser sensor, by obtaining the section wheel of rail 8 after demarcation Wide sized data, the measurement of profile of steel rail parameter is carried out using the data.
The device includes four sensors, sensor stand 5, base 6, orbit transports roller 7 and rail 8, wherein:Four biographies Sensor is laser profile sensor, is respectively first sensor 1, second sensor 2,3rd sensor 3, the 4th sensor 4, Using the linear laser distance measuring sensor (such as Gocator2350 sensors) of same model, they are separately mounted to sensor branch At four interior angles of frame 5, and the Z axis of each sensor coordinate system intersect at the center of sensor stand, and sensor stand is solid It is scheduled on the base 6 of the detection means, while orbit transports roller 7 is fixed with base, for supporting and conveying rail 8.It is described Four laser profile sensors, the profiled outline data point that it is collected makes respectively in respective sensor coordinate system Data scaling must be carried out before carrying out profile of steel rail detection with the device so that the device can export complete profile of steel rail.
The present invention is comprised the following steps during it must carry out data scaling before carrying out profile of steel rail detection:
1. it is placed on detection station using the calibrated bolck of intended size, opens laser sensor;
2. the data coordinate system by four laser sensor outputs carries out rotation transformation, is unified to the same world to sit In mark system, calibrating parameters 1 are obtained, it is anglec of rotation R1, R2, R3, R4;
3. postrotational data are carried out into coordinate translation according to the size of calibrated bolck, obtain calibrating parameters 2, it is inclined data Shifting amount S1, S2, S3, S4.
The profile of steel rail detection method based on image procossing that the present invention is provided, is first demarcated with calibrated bolck, calculates rotation Gyration R1, R2, R3, R4 and translational movement S1, S2, S3, S4, later each measurement is with regard to without being demarcated again, improve inspection Survey efficiency.Rail section is made up of straight line and circular arc, and line part cannot be collected for some sensors when making rotation transformation Point, amount of calculation is increased, and standard rail is higher to its required precision as calibrated bolck, and cost increases, therefore the present invention Using rectangular block as calibrating block, its linearity is good, is easy to calculate in calibration process.
In calibration process, as shown in Figure 2 and Figure 3, first four data coordinate systems of sensor are carried out at rotation transformation Reason, then unifies in same world coordinate system to be demarcated.
The rotation transformation treatment comprises the following steps method:
Step 1:Mark the starting point of gathered data, the top margin of the size criterion block according to calibrated bolck, base and side Side;
Step 2:The top margin of acquisition or base are carried out into linear fit:
Understanding the base of sensor 1 and sensor 3 carries out linear fit, and the top margin of sensor 2 and sensor 4 is carried out linearly Fitting, because the X-direction precision of the sensor is 0.15mm, takes end points at least 5mm of the data on top margin or base away from side, The invention example takes the series data respectively given up after 50 points away from end points and flex point and makees fitting a straight line;
Step 3:Obtain four slope of fitting a straight line respectively K1、K2、K3、K4, then obtain the inclination angle of fitting a straight line Degree θ=tan-1K, angle of inclination is designated as respectively:θ1, θ2, θ3, θ4
The relative position on base or top margin and side according to each image, the anglec of rotation is respectively R1=180 ° of-θ1, R2 =180 ° of-θ2, R3=-θ3, R4=-θ4, the anglec of rotation is substituted into coordinate rotation formula respectively:
|x′i=xoffset+xi×cosR-yi×sin R
y′i=yoffset+xi×sin R+yi×cos R
In formula:R is the anglec of rotation, xoffsetIt is the translational movement of X-direction, yoffsetIt is the translational movement of Y-direction, (xi, yi) it is biography The initial data point coordinates that sensor is collected, (x 'i, y 'i) it is coordinate (xi, yi) rotated with anglec of rotation R, and with vector (xoffset, yoffset) point coordinates after translation.
Step 4:The image after every group of data rotation is obtained, and same world coordinate system is arrived into the unification of postrotational data In.
Described calibration process is to be translated four coordinate systems according to acquired image, and specific practice is to sense The image of device 1 keeps constant, and the image after its excess-three sensor rotation is translated to the image of sensor 1, its splicing Schematic diagram afterwards is as shown in Figure 4.If a length of a of calibrated bolck, a width of b, translation treatment comprises the following steps method:
Step 1:Knowable to analysis chart 3, the flex point of sensor 2 can be moved to and be overlapped with the X-coordinate of the flex point of sensor 1, Z Coordinate is the point coincidence of calibrated bolck distance wide in the upward length of Z coordinate of the flex point of sensor 1, is translated with the translational movement and sensed The view data of device 2;The flex point of sensor 3 is moved to and is overlapped with the Z coordinate of the flex point of sensor 1, X-coordinate is in sensor 1 Flex point X-coordinate to the right length for standard block length distance point overlap, with the translational movement translate sensor 3 picture number According to;The flex point of sensor 4 moves to the X-coordinate of the flex point of sensor 1 length is the distance of standard block length, sensor to the right The upward length of Z coordinate of 1 flex point is that the point of calibrated bolck distance wide overlaps, and the picture number of sensor 4 is translated with the translational movement According to;
Step 2:The fitting a straight line of the side that each sensor collects is sought with the method for linear fit in above-mentioned calibration process, The friendship of the top margin of the initial data that each sensor is collected or the fitting a straight line on base and the fitting a straight line of side is calculated respectively Point, and the point and the corresponding anglec of rotation are substituted into coordinate rotation formula, obtain postrotational flex point Zmax1, Zmax2, Zmax3, Zmax4
Step 3:The initial data that each sensor is collected carries out multi collect, gathers n times, takes out N group data By the flex point Z of step 2 gainedmax1, Zmax2, Zmax3, Zmax4, its average value is sought respectively, obtain Avgmax1, Avgmax2, Avgmax3, Avgmax4
Step 4:The translational movement of data is after the rotation of sensor 1:S1=0;The translational movement of data after the rotation of sensor 2 For:S2=Avgmax2(x,z)-Avgmax1(x,z+b);The translational movement of data is after the rotation of sensor 3:S3=Avgmax3(x,z)- Avgmax1(x+a, z), the translational movement of data is after the rotation of sensor 4:S4=Avgmax4(x,z)-Avgmax1(x+a,z+b);
Described calibration process, using a length of a=159.12mm of cross section parameter, a width of b=157.80mm, dimensional tolerance is The calibrated bolck of ± 0.02mm, by after rotation and translation conversion, being averaged through repeatedly conversion, acquired results are as shown in table 1:Pass Theoretical install angle of the sensor on sensor stand is 45 °, and as known from Table 1, the anglec of rotation is not fully with the angle of X-axis 45 °, because in the presence of manufacture alignment error, the automatic benefit to manufacture and rigging error can be realized during system calibrating Repay.
As shown in figure 5, being rail profile profile schematic diagram, rail to be detected is placed on into sensor can survey in region, use Calibrating parameters of the anglec of rotation and translational movement of demarcation to the profile of steel rail data for collecting in table 1 carry out rotation and translation change Get rail profile integrity profile in return.
By its profile size of this rail profile profile measurement, including:Rail is high, rail head is wide, the web of the rail is thick, flange of rail edge thick Degree, the flange of rail are wide, section degree of asymmetry, according to TB/T3276-2011, TB/T2344-2012 profile Size calculation principle, by split After the completion of the removal of rail profile profile repeat to obtain the region of data, with cubic spline curve piecewise fitting profile data, Profile size is calculated on SPL, acquired results are as shown in table 2.As shown in Table 2, the profile of steel rail that should be based on image procossing is examined Preferably, certainty of measurement is above ± 0.03 to the repeatability of survey method with repeatable accuracy.
The calibration result tables of data of table 1
The anglec of rotation (°) X-axis side-play amount (mm) Z axis side-play amount (mm)
Sensor 1 135.0596 0 0
Sensor 2 44.4173 1.2087 -141.7731
Sensor 3 -135.0056 150.9700 5.5178
Sensor 4 -45.2616 154.1022 -141.3155
Profile of steel rail dimension measurement result (the unit of table 2:mm)

Claims (10)

1. a kind of profile of steel rail detection means based on image procossing, it is characterized in that including four sensors, sensor stands (5), base (6), orbit transports roller (7) and rail (8), wherein:Four sensors are laser profile sensor, and they pacify respectively At four interior angles of sensor stand (5), and the Z axis of each sensor coordinate system intersect at sensor stand The heart, sensor stand is fixed on the base of the detection means (6), while orbit transports roller (7) is fixed with base, for propping up Support and conveying rail (8).
2. the profile of steel rail detection means based on image procossing according to claim 1, it is characterized in that four described biographies Sensor, using the linear laser distance measuring sensor of same model.
3. a kind of profile of steel rail detection method based on image procossing, it is characterized in that using non-contact laser image-forming principle, using Four three-dimensional laser profile sensors gather rail profile outline data, enter by the view data that each sensor is collected Row coordinate transform, rotation, translation, carry out pattern conjunction, obtain rail profile profile, the profiled outline dimensioning that will be calculated It is very little to be compared judgement with iron mark.
4. profile of steel rail detection method according to claim 3, it is characterized in that carrying out profile of steel rail parameter using the data Measurement before, first carry out data scaling as the calibrated bolck of rail profile contour detecting with rectangular block, calculate anglec of rotation R1, R2, R3, R4 and translational movement S1, S2, S3, S4, later each measurement by rail with regard to that without being demarcated again, only need to be placed on survey On amount station, directly using calibrating parameters:The anglec of rotation and translational movement obtain complete rail profile profile, wide to improve rail Shape detection efficiency.
5. profile of steel rail detection method according to claim 4, it is characterized in that using the preferable cuboid standard of the linearity Block is demarcated to it, specifically:Four sensors are symmetrically mounted on axis in 45 ° of positions, cuboid calibrated bolck Centre-height is symmetrically installed the foursquare center to be formed and is highly consistent with four sensors, then by four sensings The image that device is collected carries out Rotating Transition of Coordinate treatment, unified in same world coordinate system, is made with one of image On the basis of, the characteristic point in four images is found out, its excess-three image is translated to benchmark according to characteristic point, carry out figure spelling Close, obtain calibrated bolck profiled outline, obtain nominal data:The anglec of rotation and translational movement.
6. profile of steel rail detection method according to claim 5, it is characterized in that during data scaling is carried out, including with Lower step:
(1) it is placed on detection station using the calibrated bolck of intended size, opens laser sensor;
(2) data coordinate system by four laser sensor outputs carries out rotation transformation, is unified to same world coordinates In system, calibrating parameters 1 are obtained, it is anglec of rotation R1, R2, R3, R4;
(3) postrotational data are carried out into coordinate translation according to the size of calibrated bolck, obtains calibrating parameters 2, it is data translation Amount S1, S2, S3, S4.
7. profile of steel rail detection method according to claim 6, it is characterised in that described rotation transformation treatment include with Lower step method:
Step 1, marks the starting point of gathered data, the top margin of the size criterion block according to calibrated bolck, base and side;
Step 2, linear fit is carried out by the top margin of acquisition or base:
In four laser sensors, the base of first sensor and 3rd sensor carries out linear fit, second sensor and The top margin of four sensors carries out linear fit, because four X-direction precision of sensor are 0.15mm, takes top margin or base End points at least 5mm of the data away from side, make fitting a straight line to take the series data respectively given up after 50 points away from end points and flex point;
Step 3:Obtain four slope of fitting a straight line respectively K1、K2、K3、K4, then obtain the tilt angle theta of fitting a straight line= tan-1K, angle of inclination is designated as respectively:θ1, θ2, θ3, θ4
The relative position on base or top margin and side according to each image, the anglec of rotation is respectively R1=180 ° of-θ1, R2= 180°-θ2, R3=-θ3, R4=-θ4, the anglec of rotation is substituted into coordinate rotation formula respectively:
|x′i=xoffset+xi×cos R-yi×sin R
y′i=yoffsst+xi×sin R+yi×cos R
In formula:R is the anglec of rotation, xoffsetIt is the translational movement of X-direction, yoffsetIt is the translational movement of Y-direction, (xi, yi) it is sensor The initial data point coordinates for collecting, (x 'i, y 'i) it is coordinate (xi, yi) rotated with anglec of rotation R, and with vector (xoffset, yoffset) point coordinates after translation, before data translational movement is calculated, (x in formulaoffset, yoffset)=(0,0);
Step 4:The image after every group of data rotation is obtained, and by postrotational data unification to same world coordinate system.
8. profile of steel rail detection method according to claim 6, it is characterized in that according to postrotational image by four coordinates System carries out translation treatment, an image for sensor specifically is kept into constant, by the image after its excess-three sensor rotation Translated to the image of first sensor, obtained spliced calibrated bolck profiled outline figure, so that obtain calibrating parameters 2, i.e., Data translational movement S1, S2, S3, S4;
The translation treatment is comprised the following steps:
Step 1:The flex point of second sensor is moved to and is overlapped with the X-coordinate of the flex point of first sensor, Z coordinate is passed first The upward length of Z coordinate of the flex point of sensor is that the point of calibrated bolck distance wide overlaps, with translational movement translation second sensor View data;The flex point of 3rd sensor is moved to and is overlapped with the Z coordinate of the flex point of first sensor, X-coordinate is passed first Length is the point coincidence of the distance of standard block length to the X-coordinate of the flex point of sensor to the right, with translational movement translation 3rd sensor View data;The flex point of the 4th sensor moves to the X-coordinate of the flex point of first sensor length is standard block length to the right Distance, the upward length of Z coordinate of the flex point of first sensor overlapped for the point of calibrated bolck distance wide, translated with the translational movement The view data of the 4th sensor;
Step 2:The fitting a straight line of the side that each sensor collects is sought with the method for linear fit in above-mentioned calibration process, respectively The intersection point of the top margin of the initial data that each sensor is collected or the fitting a straight line on base and the fitting a straight line of side is calculated, and The point and the corresponding anglec of rotation are substituted into coordinate rotation formula, postrotational flex point Z is obtainedmax1, Zmax2, Zmax3, Zmax4
Step 3:The initial data that each sensor is collected carries out multi collect, gathers n times, takes out the process of N group data The flex point Z of step 2 gainedmax1, Zmax2, Zmax3, Zmax4, its average value is sought respectively, obtain Avgmax1, Avgmax2, Avgmax3, Avgmax4
Step 4:The translational movement of data is S after the rotation of first sensor1=0, the translational movement of data after the rotation of second sensor It is S2=Avgmax2(x,z)-Avgmax1(x, z+b), the translational movement of data is S after the rotation of 3rd sensor3=Avgmax3(x, z)-Avgmax1(x+a, z), the translational movement of data is S after the rotation of the 4th sensor4=Avgmax4(x,z)-Avgmax1(x+a,z+ b)。
9. in claim 3 to 7 any methods described application, it is characterized in that the method by this rail profile profile measurement its Application in profile size, the profile size includes that rail is high, rail head is wide, the web of the rail is thick, flange of rail edge thickness, the flange of rail are wide, section Degree of asymmetry.
10. application according to claim 9, it is characterized in that wide according to GB TB/T3276-2011, TB/T2344-2012 Shape size specifies Computing Principle, the rail profile profile removal after the completion of split is repeated to obtain the region of data, with three samples Bar curve segmentation be fitted profile data, on SPL calculate profile size, acquired results repeatability preferably, certainty of measurement with Repeatable accuracy is above ± 0.03.
CN201710147303.XA 2017-03-13 2017-03-13 A kind of profile of steel rail detection device and method based on image procossing Active CN106840033B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710147303.XA CN106840033B (en) 2017-03-13 2017-03-13 A kind of profile of steel rail detection device and method based on image procossing

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710147303.XA CN106840033B (en) 2017-03-13 2017-03-13 A kind of profile of steel rail detection device and method based on image procossing

Publications (2)

Publication Number Publication Date
CN106840033A true CN106840033A (en) 2017-06-13
CN106840033B CN106840033B (en) 2019-11-26

Family

ID=59144196

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710147303.XA Active CN106840033B (en) 2017-03-13 2017-03-13 A kind of profile of steel rail detection device and method based on image procossing

Country Status (1)

Country Link
CN (1) CN106840033B (en)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107632022A (en) * 2017-08-30 2018-01-26 武汉理工大学 A kind of detection method of surface flaw of steel rail and device based on data processing
CN107717235A (en) * 2017-11-27 2018-02-23 江苏昆太工业装备有限公司 A kind of product piece laser marking technique
CN108489383A (en) * 2018-03-01 2018-09-04 北京科技大学 A kind of measuring device and method of H-type cross dimensions
CN108534836A (en) * 2018-05-11 2018-09-14 武汉理工大学 A kind of automatic detection device and method for cylindrical rock sample
CN108917688A (en) * 2018-07-27 2018-11-30 武汉理工大学 It is a kind of to eliminate the hole edge distance measurement method that roll shape beam planar distortion influences based on straight line fitting
CN110132990A (en) * 2019-06-15 2019-08-16 梁帆 A kind of profile of steel rail detection method based on image recognition
CN110143218A (en) * 2019-06-27 2019-08-20 合肥工业大学 A kind of high-speed railway track switch straightness and distortion angle detection method and its detection device
CN110986817A (en) * 2019-11-22 2020-04-10 北京交通大学 Device and method for measuring initial alignment of temporary component

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140029018A1 (en) * 2011-01-25 2014-01-30 Data M Sheet Metal Solutions Gmbh Calibration of laser light section sensors during simultaneous measurement
CN104180767A (en) * 2013-05-24 2014-12-03 武汉铁路局武汉大型养路机械运用检修段 steel-rail wear measurement device based on machine vision and grinding strategy method thereof
CN104228874A (en) * 2014-09-28 2014-12-24 中南大学 Rail vehicle exterior outline noncontact detection system
CN104266608A (en) * 2014-10-22 2015-01-07 河北科技大学 Field calibration device for visual sensor and calibration method
CN104535017A (en) * 2014-12-19 2015-04-22 合肥市百胜科技发展股份有限公司 Online contour measuring device
CN104567728A (en) * 2014-12-24 2015-04-29 天津大学 Laser vision profile measurement system, measurement method and three-dimensional target

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140029018A1 (en) * 2011-01-25 2014-01-30 Data M Sheet Metal Solutions Gmbh Calibration of laser light section sensors during simultaneous measurement
CN104180767A (en) * 2013-05-24 2014-12-03 武汉铁路局武汉大型养路机械运用检修段 steel-rail wear measurement device based on machine vision and grinding strategy method thereof
CN104228874A (en) * 2014-09-28 2014-12-24 中南大学 Rail vehicle exterior outline noncontact detection system
CN104266608A (en) * 2014-10-22 2015-01-07 河北科技大学 Field calibration device for visual sensor and calibration method
CN104535017A (en) * 2014-12-19 2015-04-22 合肥市百胜科技发展股份有限公司 Online contour measuring device
CN104567728A (en) * 2014-12-24 2015-04-29 天津大学 Laser vision profile measurement system, measurement method and three-dimensional target

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
鞠标等: "基于几何特征的钢轨磨耗检测系统的研究", 《铁道标准设计》 *

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107632022A (en) * 2017-08-30 2018-01-26 武汉理工大学 A kind of detection method of surface flaw of steel rail and device based on data processing
CN107717235A (en) * 2017-11-27 2018-02-23 江苏昆太工业装备有限公司 A kind of product piece laser marking technique
CN108489383A (en) * 2018-03-01 2018-09-04 北京科技大学 A kind of measuring device and method of H-type cross dimensions
CN108534836A (en) * 2018-05-11 2018-09-14 武汉理工大学 A kind of automatic detection device and method for cylindrical rock sample
CN108534836B (en) * 2018-05-11 2020-01-14 武汉理工大学 Automatic detection device and method for cylindrical rock sample
CN108917688A (en) * 2018-07-27 2018-11-30 武汉理工大学 It is a kind of to eliminate the hole edge distance measurement method that roll shape beam planar distortion influences based on straight line fitting
CN110132990A (en) * 2019-06-15 2019-08-16 梁帆 A kind of profile of steel rail detection method based on image recognition
CN110143218A (en) * 2019-06-27 2019-08-20 合肥工业大学 A kind of high-speed railway track switch straightness and distortion angle detection method and its detection device
CN110986817A (en) * 2019-11-22 2020-04-10 北京交通大学 Device and method for measuring initial alignment of temporary component
CN110986817B (en) * 2019-11-22 2021-03-09 北京交通大学 Method for measuring initial linear shape of temporary component

Also Published As

Publication number Publication date
CN106840033B (en) 2019-11-26

Similar Documents

Publication Publication Date Title
CN106840033A (en) A kind of profile of steel rail detection means and method based on image procossing
US8078025B2 (en) Vehicle dynamic measurement device and method for comprehensive parameters of rail wear
CN105674896B (en) Contact net geometric parameter dynamic testing method based on triangulation
CN107664483B (en) A kind of cylinder bar shape parameter measurement method
CN200955949Y (en) Ceramic tile size and shape on-line vision measuring system
CN107121093A (en) A kind of gear measurement device and measuring method based on active vision
CN104833317A (en) Medium or heavy steel plate morphology detection system based on controllable symmetrical double-line laser angle and method thereof
CN106152955B (en) A kind of large-size axis parts detection means and method
CN208795188U (en) A kind of structured light binocular vision detection system
CN106996748A (en) A kind of wheel footpath measuring method based on binocular vision
CN104239904A (en) Non-contact detection method for external outline of railway vehicle
CN110039374A (en) A kind of visual identity guidance system, apparatus and method
KR101202320B1 (en) Instrumentation system using alignment scope and method for determining system parameters of alignment scope
CN109443214A (en) A kind of scaling method of structured light three-dimensional vision, device and measurement method, device
CN107292926A (en) Crusing robot movement locus verticality measuring method based on many image sequences
CN104228874A (en) Rail vehicle exterior outline noncontact detection system
CN110148180A (en) A kind of laser radar and camera fusing device and scaling method
CN202827631U (en) Rail detection device
CN109855531A (en) Size Measuring System and its measurement method for large format template material
CN107804708A (en) A kind of pivot localization method of placement equipment feeding rotary shaft
CN106441234B (en) Detect scaling method in a kind of 3D machine vision space
CN204730813U (en) A kind of medium plate Shap feature detection system controlled based on symmetric double line laser angle
CN208042989U (en) A kind of large-scale sheet metal works almost T-stable automatic detection device
CN108819980B (en) Device and method for online dynamic measurement of geometric parameters of train wheels
CN105205806A (en) Machine vision based precision compensation method

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