JPH0972851A - Method and apparatus for detecting weld bead of steel pipe - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect a welding bead with high accuracy by a method wherein image data obtained by utilizing an optical cutting method is preprocessed so as to be corrected to a precise image and the existence of the welding bead is judged by an arc adaptation method on the basis of the image. SOLUTION: A belt-shaped beam of light with which a steel pipe 1 to be detected has been irradiated or a light spot which is scanned is captured by an ITV camera 3, its image signal is transmitted to an image-processing and computing detection device 4, and image data is preprocessed so as to remove a noise or to restore a lacked part in an image-data preprocessing part with reference to the capered image data. Then, the image signal is corrected and processed to a correct image, and the difference between an actual image and an arc image is found by an arc adaptation method on the basis of an image. When data on the difference exceeds a preset threshold value, the data is judged to be a welding bead, and the coordinates of a part exceeding the preset value are regarded as the position of the bead.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention mainly relates to a method for detecting a welded portion of a steel pipe by an image processing method, and more specifically, a quality inspection of the welded portion (a portion where a weld bead is present on a base metal) or its inspection. TECHNICAL FIELD The present invention relates to a weld bead detection technique necessary for automating the work of positioning a weld bead portion in a specific direction for the work of reworking a portion.

[0002]

2. Description of the Related Art As a method for detecting a weld bead portion of a steel pipe,
It is roughly divided into an optical feature extraction method and a displacement measurement method,
Among them, the optical feature extraction method is classified into a non-contact type, the displacement measurement method is classified into a contact type and a non-contact type, and the non-contact type of the optical feature extraction method detects the brightness of the pipe surface and A brightness change detection method for extracting a brightness change specific to a part and recognizing it as a bead (Japanese Patent Laid-Open No. 5-18904 and Sho 59-2).
3203), and irradiate a laser slit light on the pipe surface to detect a tilt change from a surface shape change,
There is a tilt image detection method that recognizes a portion with a large change as a bead. On the other hand, the displacement measuring contact type uses a touch roller and an eddy current type distance sensor together to detect the bead part by following the change in the pipe shape, and the runout of the outer weld of the running pipe There is a contact roller type (see Japanese Examined Patent Publication No. 59-2593) that is used for the control to eliminate the runout by detecting with the eddy current type distance sensor, and a plurality of exciting coils and detecting coils are used. Place and
A method for recognizing a bead part from a detected waveform (Japanese Patent Laid-Open No. 5-13
3940 gazette)).

[0003]

However, since the brightness change detection method is mainly based on the method of distinguishing the brightness of the bead portion and the other portion, it is difficult to distinguish the bead portion and the other portion depending on their properties. There is. For example, when the bead portion is rusted and exhibits the same properties as the base metal portion, it is impossible to specify and detect the weld bead. In addition, the tilt image detection method can irradiate the laser slit light obliquely to the bead, so even a low bead is amplified to a high change,
The detection rate is higher than other methods. However, there is a drawback that a bead having a smooth shape cannot be accurately detected in principle.

On the other hand, in the case of the contact type in which the touch roller and the eddy current type distance sensor are used together, the distance detection characteristic is greatly changed due to the change in the magnetic permeability due to the difference in the welding characteristic (temperature, material etc.) and the pipe size (especially the plate thickness). However, there is a drawback that it is extremely difficult to achieve a high detection rate. There is also the same contact type, called a welding point sensor,
This is to recognize the change in magnetic permeability of the welded part and recognize it as a bead part.However, since the detection characteristics vary greatly depending on the change in pipe diameter and the distance of surface irregularities, it should be used for applications requiring a high detection rate. I can't. Further, the contact roller type is difficult to perform stable detection when the bead is very smooth and low, or when the bead shape varies, and is not suitable for bead detection of a randomly rotating pipe.
In the case of arranging multiple excitation coils and detection coils and recognizing the bead part from the detected waveform, it is easy to operate due to subtle unevenness on the surface of the non-detection object, and it is difficult to avoid erroneous detection. Is not suitable for the bead detection of pipes that change randomly and rotate randomly.

In addition to these detection methods, there is also a method of detecting the bead position by measuring the profile of the steel pipe surface with a contact or optical sensor and processing the profile data. However, with this method, it was difficult to detect the convex portion of the weld bead of the steel pipe having a large product and low roundness. In the case of processing profile data, the processing method is set in the expectation that the bead shape will change from the smooth arc shape of the base metal to a considerably sharp gradient. , The edge of the bead (the boundary between the base metal and the bead) changes smoothly as a profile,
Compared with other base metal parts, no particular sudden change in the slope is observed. Therefore, under such circumstances, it is difficult to distinguish between the bead and the base metal, and the range of the bead cannot be specified. Therefore, the conventional sensor that processes the profile so far cannot be applied to the bead detection. .

As another method of the prior art, the surface of the steel pipe is irradiated with fan-shaped light or the light spot is scanned on the surface of the steel pipe, and the strip-shaped light beam of the fan-shaped light irradiated on the surface of the steel pipe or the light spot to be scanned is detected. A method of measuring the profile of the steel pipe surface has also been attempted by utilizing the light cutting method. This method
From the profile captured as image information, the method is similar to the above method, in which it is expected that the bead portion will have a considerably sharp gradient change from the smooth arc shape of the base material. Similar to, it is difficult to apply in a situation where the end of the bead has a very smooth profile.

The present invention has been made in view of such problems of the prior art, and can be applied even in a situation where the end portion of the bead draws an extremely smooth profile due to the progress of welding technology in recent years and the accuracy thereof. An object of the present invention is to propose a weld bead detection method for a steel pipe and a weld bead detection device capable of well detecting a bead.

[0008]

A welding bead detection method for a steel pipe according to the present invention is directed to irradiating the surface of the steel pipe with a fan-shaped light or scanning a light spot on the surface of the steel pipe, and irradiating the surface of the steel pipe with a strip-shaped beam of fan-shaped light. Or it is a method to detect the position of the weld bead on the surface of the steel pipe by applying the arc fitting method to the image data obtained by using the light section method to detect the light spot to be scanned. While rotating in the circumferential direction, the steel pipe surface is irradiated with a fan-shaped light or a light spot is scanned on the steel pipe surface, and a strip-shaped light beam or a scanned light spot of the fan-shaped light irradiated on the steel pipe surface is detected as an image signal, After performing noise removal processing of the detected image signal, restoration of the missing portion, image processing of inclination correction, the image processing data and arc data obtained by applying the previously detected arc of the steel pipe outer circumference circle to the data. And the difference Therefore, when the difference data exceeds a preset bead detection threshold value, it is determined that there is a convex portion corresponding to the weld bead, and the width of the range exceeding the threshold value is further calculated. Is determined to be a weld bead when the bead width includes a preset allowable range, and the position of the protrusion is calculated as a weld bead position.

That is, according to the method of the present invention, the image data obtained by utilizing the optical cutting method is not used as it is, but it is possible to accurately detect even a low and smooth weld bead on the surface of the steel pipe. Pre-processing of image data such as noise removal and restoration of missing parts is performed on the image data. The reason for this is that images captured at the production site usually have non-uniformity on the surface of the steel pipe, and their profile data often have missing data or contain information containing noise components. As a result, the presence or absence of the bead and the position of the bead cannot be accurately known as a cause of erroneous detection. The missing portion of the data is repaired by calculating a line segment connecting the coordinates of the two pieces of data on both sides of the missing data and filling the missing portion with the data along this line segment. Further, since the roundness of the steel pipe is not sufficient, when such a steel pipe is rotated in the circumferential direction, the profile detected every moment becomes an inclined profile, which requires correction. This profile correction method can be easily performed by a calculation formula.

Next, when the arc of the steel pipe outer circumference circle detected in advance by the arc fitting method is applied to the accurate image processing data thus obtained, the arc data of the steel pipe outer circumference circle is applied. When a portion where the image processing data (actual image) overflows occurs, the difference between the portions is calculated. From this difference data, it is determined whether or not it is a convex portion corresponding to a welding beat. Whether or not the protrusion corresponds to the weld bead is compared with a preset threshold value (allowable value), and when the difference data exceeds the threshold value, the protrusion corresponding to the weld bead. It is determined that there is, and if it is less than the set value, it is determined that there is no convex portion corresponding to the welding bead. Then, as a result of the determination, when the presence of a convex portion corresponding to the weld bead is confirmed, the width of the convex portion is calculated and obtained, and when the width value matches the preset width of the bead, Is determined to exist, and the coordinates of the central portion of the range exceeding the set value are used as the bead position.

Further, a bead detection device for carrying out this method is a light source for irradiating the surface of a steel pipe rotating in the circumferential direction with a fan-shaped light, or an optical scanner for scanning a light spot on the steel pipe surface, and a steel pipe surface. An image detector for detecting, as an image signal, a strip-shaped ray of the fan-shaped light or a light spot scanned on the surface of the steel pipe, and an image for detecting a welding bead based on the image signal detected by the detector. And a processing operation detecting device, wherein the image processing operation detecting device includes an image data capturing section for capturing the image signal detected by the image detector, and noise removal of the image data captured by the device. processing,
An image data pre-processing unit that performs image processing for repairing missing portions and correcting inclination, an arc fitting unit that fits an arc of the steel pipe outer circumference circle detected in advance to the image data pre-processed by the device, and the pre-processed image A difference calculation unit for obtaining the difference between the data and the arc data obtained by applying the arc, and corresponds to a welding bead when the difference data obtained by the device exceeds a preset bead detection threshold value. While determining that there is a convex portion, further calculating the width of the range that exceeds the threshold value, when the width matches the bead width including the preset allowable range, the convex portion is determined to be a welding bead,
And a bead position calculation unit that calculates the position of the convex portion as a welding bead position.

Image detectors include ITV cameras and PS
D or the like can be used. In the image data capturing section of the image processing calculation detecting device, noise removal processing of image data captured by the image data pre-processing section for image data obtained by an image detector such as an ITV camera and restoration of a missing portion are performed. , Pre-processing such as inclination correction is performed. Next, an arc fitting section applies an arc to the preprocessed image data to calculate difference data. The function of the arc fitting unit is to calculate the average radius and average center coordinates of the arc of the actual image, calculate the coordinates of the arc, and calculate the difference between the image data and the arc data by the difference calculation unit based on the calculated values.

The bead position calculation unit determines that a bead may exist in the screen when the difference data exceeds the threshold value for bead detection, and further calculates the width of the range exceeding the set value. However, if this width matches the preset bead width (including the allowable value), it is determined that this image contains a bead, and the center of the range of the portion exceeding the set value is determined. The coordinates are output as the bead position.

[0014]

1 is a schematic diagram showing a structural example of a welding bead detecting device for a steel pipe according to the present invention, and FIG. 2 is a block diagram showing a structure of an image processing calculation detecting device in the above welding bead detecting device. 3, FIG. 3 is a diagram showing an example of an image captured by an image data capturing unit of the image processing calculation and detection device of the above, FIG. 4 is a diagram showing an example of an image which is vertically and horizontally fluctuated, and FIG. FIG. 6 is a diagram showing a corrected image, FIG. 6 is a diagram for explaining the calculation of an arc fitting part of the image processing computation detecting device, and FIG. 7 is a diagram showing an example of a state in which the arc image is fitted to an actual image in the arc fitting part. Is an explanatory view showing an example of a method of obtaining a bead position and a difference between actual image data and arc data, 1 is a steel pipe to be detected, 1-2 is a weld bead, 2 is a light source or an optical scanner, 3 is an ITV camera, 4 Is an image processing calculation detecting device, 4-1 Image data acquisition unit, 4-2 image data pre-processing unit, 4-3 arc fitting portions, 4-4 difference computing unit, 4-5 bead position calculating unit, 5 is a monitor.

A bead detection method using the light cutting method will be described with reference to FIG. 1. The steel pipe 1 to be detected is irradiated with fan-shaped light from a light source or an optical scanner 2, or a light spot is detected on the surface of the steel pipe to be detected. The ITV camera 3 scans and captures the light beam or the light spot of the band radiated on the steel pipe 1 to be detected by the ITV camera 3.
In this method, the image signal is transmitted to the image processing calculation detecting device 4 to detect a bead.

In the image processing calculation detecting device 4, as shown in FIG. 2, first, the image data captured by the ITV camera 3 is
The image data pre-processing unit 4-2 inputs the image data input unit 4-1 for inputting the image data into the storage device in the device, and removes noise or a missing portion in the image data pre-processing unit 4-2. The preprocessing of the image data such as the restoration of is performed.

FIG. 3 shows an example of an image input to the image data capturing section 4-1. The vertical axis (Y axis) and the horizontal axis (X axis) are
The vertical and horizontal axes of the ITV camera 3 are equivalent to each other, and an arc-shaped profile showing a pipe is incorporated. However, images captured at the production site usually have non-uniformity on the surface of the steel pipe, and their profile data often contain missing data or contain noise components. In such a case, the data is removed from the analysis target. As a result, the captured data has more and more missing portions, so the image data preprocessing unit 4-2 determines the coordinates of the two data on both sides of the missing data for the missing portion of the data. An image is restored by calculating a line segment to be connected and filling the missing portion with data along the line segment.

The image data preprocessing unit 4-2 also corrects the inclination of the captured arc-shaped profile.
That is, the roundness of the steel pipe 1 to be detected may not be sufficient, and when such a steel pipe is rotated in the circumferential direction with the longitudinal direction as an axis, the profile detected momentarily greatly fluctuates from side to side and up and down. Therefore, the normally obtained profile is an inclined profile as shown in FIG. Therefore, it is necessary to correct this tilted profile. The correction method is shown below.

First, the slope coefficient α is calculated by the following equation (1). α = (Y ₂ −Y ₁ ) / (X ₂ −X ₁ ) ... Formula (1)

Next, using this inclination coefficient α, the profile data is corrected by the following equation (2). FIG. 5 shows profiles before and after the inclination correction.

PROF ′ (i) = PROF (i) + α (X ₂ −i) Equation (2) PROF ′: Profile after inclination correction PROF: Profile before inclination correction i: Values from X ₁ to X _2.

Next, the function of the circular arc fitting portion 4-3 is shown in FIG.
It will be described based on. FIG. 6 is an explanatory diagram of the arc fitting calculation of the detected steel pipe 1. First, the coordinates of the radius calculation point 1 are obtained by the following formulas (3) and (4).

X _p1 = X ₁ + (X ₂ −X ₁ ) / 4 (3) Formula X _p1 = PROF ′ (X _{p 1} ) ... (4) Formula

Similarly, the coordinates of the radius calculation point 2 are obtained by the following equations (5) and (6).

Y _p2 = X ₂ − (X ₂ −X ₁ ) / 4 (5) Formula Y _p2 = PROF ′ (X _p2 ) · (6) Formula

Next, the center of the circle is {(X ₁ + X ₂ ) /
2, Y _c1 }, the equation of a circle passing through the left end (X ₁ , Y ₁ ) of the data and the calculation point 1 (X _p1 , Y _p1 ) is obtained by the following equation (7).

R ₁ ² = {(X ₂ −X ₁ ) / 4} ² + (Y _p1 −Y _c1 ) ² = {(X ₂ −X ₁ ) / 2} ² + (Y ₁ −Y _c1 ) ² ... (7) formula

Similarly, the center of the circle is {(X ₁ + X ₂ ) /
2, Y _c ² }, the equation of a circle passing through the right end (X ₂ , Y ₂ ) of the data and the calculation point 2 (X _p2 , Y _p2 ) is obtained.

R ₂ ² = {(X ₂ −X ₁ ) / 4} ² + (Y _p ² −Y _c ² ) ² = {(X ₂ −X ₁ ) / 2} ² + (Y ₂ −Y _{c 2} ). ² ... (8) formula

The average radius and average center coordinates of the two circles obtained by the above equations are obtained by the following equations (9) to (11).

R ₀ = (r ₁ + r ₂ ) / 2 (9) Expression X ₀ = (X ₁ + X ₂ ) / 2 (10) Expression Y 0 = (Yc ₁ + Yc ₂ ) / 2 (11) Expression

From the respective values obtained by the above equations (9) to (11), the coordinates of the arc are obtained by the following equation (12). FIG. 7 is an arc fitting diagram, and shows the image data PROF ′ (i) and the arc data C (i) together.

C (i) = Y ₀ + SQRT {r0 ² − (i−X ₀ ) ² } (12) Equation (12) Here, SQRT indicates a square operation.

Next, the difference calculation unit 4-4 calculates the difference SA (i) between the image data and the arc data by the following equation (13).

SA (i) = PROF ′ (i) −C (i) (13) Formula

FIG. 8 is an enlarged view of the dotted circle portion of FIG.
The calculation result of the difference data SA (i) is shown. When this difference data exceeds a preset threshold value for beat detection, it is primarily determined that a bead may be present in the screen, and a range exceeding the threshold value ( The width W of the shaded area in the figure) is calculated, and if this width matches the preset bead width (including the allowable value),
It is finally determined that this image contains beads. For example, when using a bead specification with a width of 30 mm, as an example, 20 m
When the condition of m <detection width <40 mm is satisfied, it is judged as an official bead. The reason is that the allowable range must be set so that foreign matter such as welding spatter is not mistaken for a bead.

In FIG. 8, when the height detection resolution of the vertical axis (Y axis) is about 0.1 mm / 1 bit, the bead height is about 2 mm (20 bits). On the other hand, a change of about 5 bits can be seen in areas other than the bead portion. Therefore, the correct position cannot be detected unless the changed portion of 6 bits or more is used as a bead. In the example of FIG. 8, the change portion of 10 bits or more is set to be seen as a bead in consideration of safety. Practically, the changed portion of 6 to 7 bits or more may be determined as a bead (primary) (threshold setting).

Finally, the bead position calculation unit 4-5
At, the coordinates of the central portion of the range exceeding the threshold value (the hatched portion in the figure) are obtained and set as the bead position, and the relevant data is output.

[0039]

As described above, the present invention does not use the image data obtained by utilizing the optical cutting method as it is, but removes noise from the image data or repairs the missing portion. It is a method to pre-process the image data to correct it and to determine the presence or absence of the welding bead by the arc fitting method based on this image, so even when the welding bead part draws a smooth profile. The bead can be detected accurately without being affected by gloss. Therefore, the present invention has an extremely high bead detection rate as compared with a conventional feature extraction method by image processing that determines the presence or absence of beads by pattern change based on brightness and luminance, and quality inspection of welds, or This greatly contributes to the automation of the work of positioning the weld bead portion in a specific direction for the work of repairing the portion.

[Brief description of drawings]

FIG. 1 is a schematic view showing a device configuration example of a weld bead detection device for a steel pipe according to the present invention.

FIG. 2 is a block diagram showing a configuration of an image processing calculation detecting device in the welding bead detecting device of the above.

FIG. 3 is a diagram showing an example of an image captured by an image data capturing unit of the image processing calculation and detection apparatus of the above.

FIG. 4 is a diagram showing an example of an image that is vertically and horizontally fluctuated in an image data preprocessing unit of the image processing calculation and detection apparatus.

FIG. 5 is a diagram showing image profiles before and after inclination correction in the image data preprocessing unit.

FIG. 6 is an operation explanatory diagram of an arc fitting unit of the image processing operation detection device of the above.

FIG. 7 is a diagram showing an example of a state where an arc image is fitted to an actual image in the arc fitting unit.

FIG. 8 is an explanatory diagram showing an example of a method of obtaining a bead position and a difference between actual image data and arc data in a difference calculation unit of the image processing calculation detection device.

[Explanation of symbols]

 1 Steel Pipe to be Detected 1-2 Welding Bead 2 Light Source or Optical Scanner 3 ITV Camera 4 Image Processing Calculation Detection Device 4-1 Image Data Capture Unit 4-2 Image Data Pre-Processing Unit 4-3 Arc Fitting Unit 4-4 Difference Calculation unit 4-5 Bead position calculation unit 5 Monitor

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Naoto Harasaki, 3 Hikari, Kashima City, Kashima City, Ibaraki Prefecture Sumitomo Metal Industries, Ltd. Kashima Steel Works (72) Inventor Mamoru Kawahara 3-176 Fukusaki, Minato-ku, Osaka City, Osaka Prefecture Kansai Tech Co., Ltd. (72) Inventor Nobuyoshi Ikawa 3-1-1176, Fukusaki, Minato-ku, Osaka-shi, Osaka Kansai-Tech Co., Ltd.

Claims (2)

[Claims] 1. A steel pipe surface is irradiated with a fan-shaped light or a light spot is scanned on the steel pipe surface while rotating the steel pipe in a circumferential direction,
The strip-shaped light beam of the fan-shaped light radiated to the surface of the steel pipe or the light spot to be scanned is detected as an image signal, and noise removal processing of the detected image signal, restoration of a missing portion, image processing of tilt correction, and then the The difference between the image processing data and the arc data obtained by applying the arc of the steel pipe outer peripheral circle detected in advance to the data is obtained, and the welding bead is obtained when the difference data exceeds a preset bead detection threshold value. In addition to determining that there is a convex portion corresponding to, the width of the range that exceeds the threshold value is further calculated, and if this width matches the bead width that includes the preset allowable range, the convex portion is a welding bead. A method for detecting a weld bead in a steel pipe, which comprises determining and calculating the position of the convex portion as a weld bead position. 2. A light source for irradiating a surface of a steel pipe rotating in the circumferential direction with a fan-shaped light, or an optical scanner for scanning a light spot on the surface of the steel pipe, and a strip-shaped beam or a steel pipe of the fan-shaped light irradiated on the surface of the steel pipe. An image detector for detecting a light spot scanned on the surface as an image signal, and an image processing operation detecting device for detecting a welding bead based on the image signal detected by the detector, the image processing The calculation and detection device includes an image data capturing unit that captures the image signal detected by the image detector, a noise removal process for the image data captured by the device, an image process for correcting a missing portion, and an inclination correction. An image data pre-processing unit, an arc fitting unit that applies an arc of the steel pipe outer circumference circle detected in advance to the image data pre-processed by the apparatus, and an arc obtained by applying the pre-processed image data and the arc. data And a difference calculation unit for obtaining a difference between the difference data obtained by the apparatus and a bead detection threshold value set in advance, and it is determined that there is a protrusion corresponding to the welding bead. Calculating the width of the range that exceeds the threshold value, if this width matches the bead width that includes the preset allowable range, judge the convex portion as a welding bead and calculate the position of the convex portion as the welding bead position. A weld bead detection device for a steel pipe, comprising: