JPH0933233A - Measuring apparatus for bending of long work - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an apparatus by which the bending of a long work is measured efficiently and with good accuracy without requiring man power. SOLUTION: The measuring apparatus is provided with a long inspection base 10 on which a work W is placed. Three or more cameras S1 to S4 are installed at an interval on a straight line along the length direction of the inspection base 10. The cameras S1 to S4 take photographs of edges of the work W respectively in a plurality of places in the length direction. Image processing computers GC1 to GC4 process images by the cameras S1 to S4, and they detect the coordinates of the edges for every camera. A host computer HC computes the bending of the work W on the basis of the correlation between a plurality of edge coordinates obtained by the image processing computed GC1 to GC4.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an apparatus for measuring the degree of bending of a long workpiece such as welded H-section steel.

[0002]

2. Description of the Related Art Conventionally, this type of work bending inspection is
As shown in FIG. 11, the inspector relied on visual confirmation in the direction of arrow A. If it is difficult to confirm visually, water thread B
Was attached to both ends of the work W, and the maximum clearance ε between the water thread B and the work W was measured and compared with the standard allowable value.

[0003]

The visual inspection by an inspector is a sensory one, so that it requires skill and there is a problem that the variation is large. In addition, the method of inspecting each work with a string of water requires three or more inspectors (two people holding both ends of the water thread and one person measuring the maximum gap), which is inefficient. there were.

In view of the above circumstances, it is an object of the present invention to provide a long workpiece bending measuring apparatus capable of efficiently and accurately measuring the degree of bending of a workpiece without requiring manpower. To do.

[0005]

A bend measuring device according to the invention of claim 1 is provided with three or more cameras which are arranged on a straight line at intervals and photograph the edges of a work extending along the straight line. Image processing means for detecting the coordinates of the edge of each camera from the image information obtained by each camera, and bending calculation means for calculating the degree of bending of the workpiece based on the edge coordinate information obtained by the image processing means. And are provided.

According to a second aspect of the present invention, in the apparatus for measuring the bending of a long work according to the first aspect, an illumination means for illuminating the work from an obliquely upper side of the work is further provided, and the upper surface of the work is illuminated by the illumination means. It is characterized in that the image is brightened and a shadow is formed on the side of the work, and the camera obtains an image that is divided into bright and dark with the edge as a boundary.

According to a third aspect of the present invention, in the long workpiece bending measuring apparatus according to the first or second aspect, the image processing means sets a plurality of search lines in the image of each camera, and the search is performed. It is characterized in that the intersection of the line and the edge included in the image is detected as the coordinate of the edge, and this coordinate information is sent to the bending calculation means.

According to a fourth aspect of the present invention, in the bending measuring apparatus for a long work according to the third aspect, the image processing means obtains coordinate information of a reference line from an image of a true straight line taken by a camera. The coordinate calculation information is sent to the bending calculation means, and the bending calculation means stores the coordinate information of the reference line, and the coordinates of the edge of the actual work sent from the image processing means are used as the coordinate information of the reference line. By correcting based on this, the absolute coordinates of the edge with respect to this reference line are obtained,
It is characterized in that the bending of the work is calculated based on the absolute coordinates.

According to a fifth aspect of the present invention, in the long workpiece bending measuring apparatus according to the fourth aspect, the bending calculation means obtains a comparison straight line representing the inclination of the work from the absolute coordinates, and from the comparison straight line. It is characterized in that the comparison coordinates are calculated as the position of the absolute coordinates of, and the degree of bending of the work is calculated based on the mutual relation of the comparison coordinates.

[0010]

BEST MODE FOR CARRYING OUT THE INVENTION An embodiment of the present invention will be described below with reference to the drawings. As shown in FIGS. 1 and 2, the bending measuring apparatus includes an inspection table 10 extending linearly in the Y-axis direction.
Above the inspection table 10, four CCD cameras (any number of three or more cameras) S1 to S4 are installed at intervals along the Y-axis direction, and above the inspection table 10. , Four illuminators (illuminating means) R1 to R4, which are installed in correspondence with the cameras S1 to S4, respectively, and are separated from the inspection table 10 in the X-axis direction, and four image processing computers connected to the cameras S1 to S4, respectively. (Image processing means) GC1 to GC4 and image processing computer GC
A host computer HC (bending calculation means) that processes data from 1 to GC4 and a display 20 such as a CRT that displays the processing content of the host computer HC are provided. As shown in FIGS. 2 and 3, the cameras S1 to S4
Is facing downward, and the illuminators R1 to R4 are facing obliquely downward, that is, facing the inspection table 10.

As shown in FIG. 3, a long work W to be inspected is, for example, welded H-shaped steel after strain correction processing, and has a pair of flanges 12 and a web 11 connecting these flanges 12. ing. The work W is placed on the inspection table 10 in a posture in which both the flanges 12, 12 are vertically erected. In this state, the end faces 12a of both flanges 12
Is horizontal and looks upward. The longitudinal direction of the work W substantially coincides with the longitudinal direction of the inspection table 10, that is, the Y-axis direction.

As best shown in FIG. 3, the cameras S1 to S4 are located to the right of the center of the work W in the width direction. The illuminators R1 to R4 are located on the left of the work W,
By illuminating the surface of the work W obliquely, a shadow SW is formed to the right of the edge E of the end face 12a of the right flange 12. As a result, as shown in FIG. 4, the edge E is made to stand out as a boundary between the bright region of the illuminated end surface 12a and the dark region of the shadow SW.

With this arrangement, the cameras S1 to S4 are
In the image, the region including the edge E of the work W is photographed at a plurality of positions in the longitudinal direction of the work W, and images G1 to G4 as shown in FIG. 4 are obtained. The cameras S1 and S4 at both ends
Is at a position where both ends of the work W can be photographed.

The image processing computers GC1 to GC4 are
It has a function of detecting the coordinates of the edge E of each of the cameras S1 to S4 by processing the images obtained by the cameras S1 to S4. That is, the image processing computers GC1 to GC4, as shown in FIG.
The coordinates of the edge E (intersection between the edge E and the search line F) are detected by searching the image data G1 to G4 of S4 on a plurality of search lines F parallel to the X direction. The search line F is selected from the scanning lines in each of the cameras S1 to S4 and is spaced from each other at equal intervals.

The host computer HC has a function of calculating the degree of bending of the work W on the basis of the coordinates of the edge E obtained by the image processing computers GC1 to GC4 and determining the quality of the curve.

Before measuring the bending of the work W, as shown in FIG. 5, the water thread 1 (true straight line) is stretched in the longitudinal direction of the inspection table 10, that is, along the Y axis. The image processing computers GC1 to GC4 use the images G1 to G1 captured by the cameras S1 to S4.
In G4, the distance (the number of pixels) from one end of the search line F to the water thread 1 is detected as the coordinates of the reference line in each of the cameras S1 to S4 and sent to the host computer HC. The host computer HC stores the coordinate information of the reference line 1. It should be noted that although the same number of coordinates of the reference line 1 as each camera is obtained as the number of search lines F, one symbol δ1 to δ4 is shown for each camera in FIG. 5 in order to simplify the drawing. A laser beam or the like may be used as the reference line instead of the water thread.

In the host computer HC, when measuring the bending of the work W, the coordinates of the edge E sent from the image processing computers GC1 to GC4 are corrected based on the reference line coordinates δ1 to δ4 to correct the reference. Find the absolute coordinates for a line. A comparison straight line representing the inclination of the work W is obtained from these absolute coordinates, the comparison coordinates of the edge E with respect to the comparison straight line are obtained, and the degree of bending of the work W is calculated from the mutual relationship of the comparison coordinates.

Next, details of the bending measurement will be described. First, the processing contents of the image processing computers GC1 to GC4 will be described based on the flowchart of FIG.
When the process is started, the image data of the camera is fetched (step S1), the number of search lines F as shown in FIG. 4 is determined (step S2), and the brightness value (luminance value) on the search line F is stored ( Step S3). Here, in order to absorb the error, the sum of the brightness values of the scanning lines which are shifted by one pixel on both sides of the search line F is set as the brightness value of the search line.

Next, the difference processing, that is, the difference between the brightness values of the pixels adjacent to each other along the search line F is obtained (step S
4) From the result of the difference, the pixel position that satisfies the conditions of "the maximum value of the difference" and "the brightness decreases" is determined as the provisional coordinates of the edge E (step S5), and on the four search lines F. A virtual straight line is obtained from the four tentative coordinates by the method of least squares, and a search width within a certain range is determined for this virtual straight line (step S6). Then, of the stored brightness value data, only the data within the search width is copied to the memory area for image processing (step S7). This is to improve the efficiency of image processing.

Next, the maximum lightness value and the minimum lightness value are detected from the data copied to the memory area (step S8), and based on these maximum lightness value and the minimum lightness value, they serve as a reference for edge determination. A threshold value (boundary lightness value) slice is obtained (step S9). Then, the edge position is searched for each search line F by the threshold value slice (step S10). The edge position search is performed by sequentially comparing the brightness values of adjacent pixels with a threshold value slice, and pixel positions d1 and d2 satisfying the condition of d1 ≧ slice d2 ≦ slice are determined as the coordinates of the edge E.

When the pixel position d2 and the pixel position d1 do not match, the edge position d is obtained by the proportional calculation of the following equation (see FIG. 6). d = d1 + (d1c-slice) / (d1c-d2c) where d1c = lightness value of pixel d1 d2c = lightness value of pixel d2 With this method, at the intersection of the edge E and the search line F with a resolution less than one pixel. Some coordinates are required.

Then, the data of the coordinates of the edge E is stored (step S11), the data is displayed as necessary (step S12), the data is transferred to the host computer HC (step S13), and the process is terminated.

Next, the processing contents of the host computer HC will be described with reference to FIG. When the processing is started, the data from the image processing computers GC1 to GC4 is received (step S21) and stored in a file (step S22). The data to be stored are the coordinates of the edge E for the number of search lines × the number of cameras. It should be noted that this file stores in advance the coordinate information of the reference line for each of the cameras S1 to S4. When the coordinates of the received edge E are shown in the figure, reference numerals D11 to D14 and D21 to D2 are shown in FIG.
4, D31 to D34, and D41 to D44. The coordinates D11 to D14 are obtained by the camera S1, and similarly, the coordinates D21 to D24 are the camera S2 and D3.
1 to D34 are cameras S3, D41 to D44 are cameras S3
They were obtained in 4 respectively.

Next, the coordinates of the absolute reference line are subtracted from the coordinates of the received edge E to obtain the coordinates (absolute coordinates) of the edge E with respect to the absolute reference line, as shown in FIG. 7B (step S23). . As a result, the misalignment of the cameras S1 to S4 is absorbed. Then, the absolute coordinate of the edge E is converted from the pixel unit to the millimeter unit, and the information of the Y-axis coordinate component is added to the absolute coordinate based on the camera pitch data. (Step S24).

Next, a straight line (comparative straight line) connecting the absolute coordinate D11 corresponding to the vicinity of one end of the work W and the absolute coordinate D44 near the other end, in other words, an inclination correction formula is obtained (step S25). The comparison straight line represents the inclination of the entire work W. Next, the position of absolute coordinates (comparative coordinates) corresponding to the comparison straight line 31 is obtained. In other words, the comparison coordinate in which the tilt component is removed from the absolute coordinate is calculated (step S26). That is, FIG. 7C shows this comparison coordinate.

Next, the curved shape of the work W is determined (step S27) based on the mutual relationship of the comparison coordinates, the amount of bending is calculated (step S28), and the quality of the work W is determined, and the determination result is obtained. Display (step S29) and terminate the process.

As the curved shape, a "large curved" type as shown in FIG. 8 (a) and an "S" as shown in FIG. 8 (b) are used.
Bent "type," Bend "as shown in Figure 8 (c)
There are types, and shape determination is performed based on the distribution of edge coordinate data.

For example, as shown in FIG. 8A, when the comparison coordinates D12 to D43 between them are all on one side with respect to the straight line 33 connecting the comparison coordinates D11 and D44 at both ends, that is, all In the case of the same sign, it is determined to be a large curved shape. Alternatively, if the gradient of the comparison coordinates near both ends has a different sign (in the figure, the gradient near the left end is an upward slope, the neighborhood near the right end is a downward slope, and the direction of the gradient is different), it is possible to determine that it is a large bend. . In the case of a large bend, the distance ε1 between the straight line 33 connecting the comparison coordinates D11 and D44 at both ends and the farthest comparison coordinate (the comparison coordinate D31 in the figure) is calculated. In other words, the absolute value ε1 of the largest comparison coordinate
Ask for. When the ratio of the distance ε1 to the total length L (about 12 m) is equal to or larger than the standard value (for example, 1/1000), it is determined as a defective product.

Further, as shown in FIG. 8B (a straight line having an X coordinate of zero is indicated by the symbol L), when the gradients between the comparison coordinates in the vicinity of both ends have the same sign (in the figure, the vicinity of the left end is a downward slope to the right). (Slope, the slope near the right end is also a slope to the right, and the direction of the slope is the same)
Is determined to be an S bend. In that case, in the illustrated example, the maximum and minimum comparison coordinates D from the data D11 and D44 at both ends.
Straight lines 34 and 35 are drawn on 34 and D21, respectively. Then, the distances ε2 and ε3 to the minimum and maximum comparison coordinates with respect to the straight lines 34 and 35 are obtained, and when either of the ratios of the distances ε2 and ε3 to the total length L is the standard value 1/1000 or more, it is determined as a defective product. To do.

Further, as shown in FIG. 8C, when the line connecting the comparison coordinates is bent only near one end, and when it is a straight line at other positions, it is determined that the nose is bent. It is not necessary to judge the nose bend. As shown in FIG. 8C, a straight line connecting the Nth comparison coordinate D24 and the N + 1th coordinate D31 from one end,
The distance ε4 from the comparison coordinate D11 at one end is obtained, and when the ratio of the distance ε4 to the total length L is the standard value 1/1000 or more, it is determined as a defective product. This quality determination is performed on both ends, and also when there is a large bend or an S bend.

In the measuring device constructed as described above, the bending can be measured efficiently and accurately without requiring any manpower, and an accurate pass / fail judgment can be made, so that 100% inspection can be performed. Also, since the measurement data can be stored in a file,
It can be used for quality control.

The number of cameras and the method of determining the curved shape can be arbitrarily changed. Further, the illuminators R1 to R4 are not necessarily required as long as a contrast image that can recognize the edges as boundaries is obtained.

[0033]

As described above, according to the first aspect of the invention, the bending of the work can be automatically and accurately measured based on the image of the camera. Therefore, it is possible to efficiently measure the bend without requiring a skilled person or the like. According to the invention of claim 2, by illuminating the workpiece obliquely, the edge of the work can be positioned at the boundary between the light and the dark so that the edge of the work can be made to stand out, so that the accurate coordinates of the edge can be detected.
As a result, the bend can be accurately measured. According to the invention of claim 3, by setting a plurality of search lines for each camera, it is possible to increase the information of the edge coordinates.
The bend can be measured with high accuracy. According to the invention of claim 4, the coordinate information of the edge of the work can be corrected on the basis of the coordinate information of the reference line even if the cameras are not accurately arranged on a straight line, so that accurate bend measurement can be ensured. can do. According to the invention of claim 5, by correcting the coordinate of the edge with the comparison straight line representing the inclination of the work, the calculation process of the bending measurement can be simplified.

[Brief description of drawings]

FIG. 1 is a plan view of a bend measuring device that constitutes an embodiment of the present invention.

FIG. 2 is a side view of the device.

FIG. 3 is a front view showing a relationship between a camera, an illuminator, and a work in the same device.

FIG. 4 is a diagram showing an image obtained by a camera of the same apparatus.

FIG. 5 is an explanatory diagram showing coordinates of a reference line for each camera in the same apparatus.

FIG. 6 is a diagram illustrating a method of edge search based on brightness in the same apparatus.

7A and 7B are diagrams showing edge coordinate information detected by the device, FIG. 7A is a diagram showing raw coordinate information transferred to a host computer, and FIG. 7B is a diagram showing absolute coordinates with respect to a reference line; (C) is a figure which shows the comparison coordinate which removed the inclination component.

FIG. 8 is a diagram showing a shape determination pattern of the apparatus.

FIG. 9 is a flowchart showing the processing contents of the image processing computer of the apparatus.

FIG. 10 is a flowchart showing processing contents of a host computer of the apparatus.

FIG. 11 is an explanatory diagram of a conventional work bending measurement method.

[Explanation of symbols]

W work E edge S1 to S4 camera F search line R1 to R4 illuminator (illumination means) GC1 to GC4 image processing computer (image processing means) HC host computer (bend calculation means)

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Seiichi Tanaka 5-5 Yonbancho, Chiyoda-ku, Tokyo 9 Topy Kogyo Co., Ltd. (72) Ahihi Kano 5-9 Yonbancho, Chiyoda-ku, Tokyo 9 TOPY Business

Claims (5)

[Claims] 1. Three or more cameras which are arranged on a straight line at intervals and photograph the edge of a work extending along the straight line, and the edge of each camera based on the image information obtained by each camera. An image processing means for detecting coordinates, and a bending calculation means for calculating the degree of bending of the work based on the coordinate information of the edges obtained by the image processing means. Measuring device. 2. A lighting means for illuminating the work from obliquely above the work is provided, and the lighting means brightens the upper surface of the work and forms a shadow on the side of the work, and the camera demarcates the edge. An apparatus for measuring the bending of a long work piece according to claim 1, wherein an image which is divided into bright and dark is obtained. 3. The image processing means sets a plurality of search lines in the image of each camera, detects an intersection of the search line and an edge included in the image as the coordinate of the edge, and the coordinate information Is sent to the bending calculation means. 3. The bending measuring device for a long workpiece according to claim 1, wherein 4. The image processing means obtains coordinate information of a reference line from an image of a true straight line taken by a camera and sends it to the bend calculating means, and the bend calculating means receives the coordinate information of the reference line. The coordinates of the edge of the actual work sent from the image processing means are stored and corrected based on the coordinate information of the reference line to obtain the absolute coordinate of the edge with respect to the reference line. The bending measurement device for a long work according to claim 3, wherein the bending of the work is calculated based on the coordinates. 5. The bending calculation means obtains a comparison straight line representing the inclination of the work from the absolute coordinates, calculates the comparison coordinates as the position of the absolute coordinates from the comparison straight line, and by the mutual relation of the comparison coordinates, The bending measurement apparatus for a long work according to claim 4, wherein the degree of bending of the work is calculated.