JPH11166898A - Device and method for inspecting rod for flaw - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately detect any linear, shallow flaw, such as a spiral flaw (circumferentially extending flaw) on the surface of a rod, and to determine with high accuracy whether or not the rod is defective according to the size of the flaw. SOLUTION: As a rod 5 serving as a subject for inspection is rotated, the surface of the rod 5 is photographed with a one-dimensional CCD camera 21 and data about the image of its overall peripheral surface are stored in an image memory 32. A flaw pixel detecting part 34 detects those pixels which are in proximity to each other in the direction of the axis of the rod and which differ in density (difference of brightness) by more than a flaw detection threshold as flaw pixels. A density difference (detection threshold) changing part 33 for detecting flaws adjusts the detection threshold so that a predetermined number or more of flaws can be detected. A flaw pixel grouping part 35 groups flaw pixels within a predetermined pixel interval in the direction of the axis, and then groups flaw pixels within the predetermined pixel interval in the circumferential direction to extract a linear flaw. A defective judging part 36 determines whether or not the rod 5 is defective by comparing the size (length, area, etc.) of the linear flaw with a criterion for defectives.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a rod flaw inspection apparatus and a rod flaw detection apparatus for detecting flaws on the surface of a cylindrical object such as a cylinder or rod used in a shock absorber or the like by using image processing to judge pass / fail. The present invention relates to a flaw inspection method, and more particularly, to a rod flaw inspection apparatus and a rod flaw inspection method capable of detecting a spiral flaw on a rod surface.

[0002]

2. Description of the Related Art JP-A-49-129585 discloses that
A technique is described in which a laser beam is scanned along a generatrix on a slowly rotating rod, and the reflected light is received by a photodetector to photoelectrically detect a flaw on the rod surface. When receiving light, a bright-field method in which light is received in the specular reflection direction with respect to the rod surface (scratch is detected as a dark area) or a dark-field method in which light is received in a direction different from the specular reflection direction (scratch is detected as a bright light area) ) Can be adopted.

[0003] Japanese Patent Application Laid-Open No. 50-75086 discloses an image of a rod surface formed on an imaging surface of an imaging tube by a low-magnification microscope objective lens opposed to the rod, and the rod is rotated around an axis. While the objective lens is moved in the axial direction, the entire surface of the rod is spirally scanned to detect a flaw on the rod surface as a video signal from an imaging tube.

Japanese Patent Application Laid-Open No. 4-1257 / 1992 discloses that a one-dimensional CCD camera having a visual field oblique to a direction in which a linear flaw is expected to be generated on a running steel sheet is disposed to face the steel sheet.
Scanning signals for the surface of the steel sheet are sequentially output from the CCD camera at predetermined intervals, and a difference between the output scanning signals is calculated to generate a difference signal corresponding to each scan. A signal exceeding the reference value is extracted as a defect signal, and the defect signal sequentially moves in the width direction of the steel sheet at a substantially constant pitch at each scanning interval, thereby determining the continuity of the defect signal and detecting a linear flaw. A technique for detecting a linear flaw in a steel sheet is described.

[0005] Japanese Patent Application Laid-Open No. 5-180777 discloses a technique in which a cylindrical object to be inspected is rotated, image information obtained by dividing the outer peripheral surface of the object into a plurality of parts in an axial direction is taken in by a television camera, and the luminance of each divided image information is obtained. There is described an appearance inspection apparatus for a cylindrical object in which components are added and a portion where the added luminance component exceeds a predetermined threshold value is determined to be defective.

[0006]

Various techniques have been proposed for inspecting a surface to be inspected for flaws based on image information of the surface to be inspected obtained by photoelectrically converting reflected light from the surface to be inspected. However, the conventional technology detects a portion where image information (for example, density value or luminance value) of a surface to be inspected exceeds a reference value (threshold value) as a flaw, and determines based on the number of detected flaws and a flaw area. The pass / fail judgment is made.

For cylinders and rods used in shock absorbers, etc., when the surface of the rod is processed (burnishing)
In some cases, spiral scratches may be formed due to scratches on the grindstone. A concave portion and a convex portion are formed in the damaged portion. In the case of the shock absorber, if the surface of the cylinder / rod has a protrusion, the protrusion may cause a scratch or a crack in the oil seal, which may cause oil leakage. For this reason, it is necessary to detect the spiral flaw on the rod surface and make the quality judgment. However, with the conventional flaw inspection technology, it was difficult to accurately detect the spiral flaw and make the quality judgment. .

For example, when a reference value (threshold) is set low (when the detection sensitivity of a flaw is set high) so that a shallow flaw can be detected by a conventional inspection apparatus, a shallow spot-like besides a spiral flaw can be detected. All the scratches are detected, and a defect is determined based on the detection result of the shallow dot-like scratches. Therefore,
If the reference value (threshold) is set high (if the flaw detection sensitivity is set low) so that only deep point flaws can be detected, shallow spiral flaws cannot be detected.

The present invention has been made to solve such a problem, and a rod flaw inspection apparatus and a rod flaw inspection method capable of accurately detecting a spiral flaw on a rod surface and making a quality judgment. The purpose is to provide.

[0010]

In order to solve the above-mentioned problems, a rod flaw inspection apparatus according to the present invention moves over the entire circumference of a rod while relatively moving the rod and an imaging unit for imaging the peripheral surface of the rod. An imaging device for imaging the rod surface, an A / D converter for converting an image signal of the rod surface output from the imaging unit into image data, an image memory for storing A / D-converted image data, and image data A flaw pixel detection unit for detecting, as a flaw, a pixel whose density difference (luminance difference) between pixels adjacent in the rod axis direction exceeds a preset flaw detection density difference value (flaw detection threshold) based on A flaw pixel grouping unit for grouping pixels in which the distance between pixels detected as flaws is within a predetermined pixel interval; and the width, length, number of pixels of the grouped flaw pixel groups and a predetermined quality Compare with criteria Comprising a quality judgment unit for judging acceptability Te.

When the number of pixels detected as flaws does not reach the preset reference number, the flaw detection density difference value (flaw detection threshold value) is changed to a smaller value. A density difference value changing unit may be provided.

In addition, the flaw pixel detecting section is provided with two in the rod axis direction.
If the density difference (luminance difference) between the pixels separated from each other exceeds a preset density difference value for flaw detection (fault detection threshold value), the density difference gradient coefficient is set to 1 and the pixel is separated by one pixel in the rod axis direction. If the density difference between the detected pixels exceeds a preset density difference value for flaw detection, the density difference gradient coefficient is set to 2, and the flaw detection density difference value changing unit determines that the cumulative value of If the value does not reach the value, the density difference value for flaw detection may be changed to a smaller value.

According to the rod flaw inspection method of the present invention, the rod surface is imaged over the entire peripheral surface of the rod while relatively moving the rod and an imaging unit for imaging the peripheral surface of the rod, and an image of the rod surface is obtained. After the signal is converted into image data and temporarily stored in the image memory, the density difference (luminance difference) between the pixels adjacent in the rod axis direction is a preset density difference value for flaw detection (flaw detection threshold value). Pixels that exceed the threshold value are detected as flaws, and then, pixels that are detected as flaws within a predetermined pixel interval are grouped to detect linear flaws. The quality is determined by comparing the width, length, and area of the flaw with a predetermined quality determination standard.

If the number of pixels detected as flaws does not reach the preset reference number, the flaw detection density difference value (flaw detection threshold value) is changed to a smaller value. The flaw detection sensitivity may be changed.

Further, when the density difference between the pixels separated by two pixels in the rod axis direction exceeds a preset density difference value for flaw detection, the density difference gradient coefficient is set to 1, and the pixels separated by one pixel in the rod axis direction are set. If the density difference between the two exceeds the preset density difference value for flaw detection, the density difference slope coefficient is set to 2, and if the cumulative value of the density difference coefficient does not reach the preset reference cumulative value, The flaw detection sensitivity may be changed by changing the density difference value (flaw detection threshold value) to a smaller value.

In the rod flaw inspection apparatus and the rod flaw inspection method according to the present invention, the determination of a flaw is made based on the density difference (luminance difference) between pixels adjacent in the rod axis direction. Even when the density of the detected image changes due to a change in the intensity of the illumination light or the like, the flaw can be accurately detected. Further, the flaw can be accurately detected without correcting the darkening of the image around the visual field range of the imaging unit (so-called shading correction).

The pixels whose intervals between pixels detected as flaws are within a predetermined pixel interval based on the density difference (luminance difference) in the rod axis direction are grouped. Wounds can be extracted.
Then, a pass / fail judgment is made based on the width, length, and area of the detected linear scratch. Therefore, even if there is a shallow point-like flaw on the rod surface, the point-like flaw and the linear flaw can be distinguished, and accurate pass / fail judgment can be made based on the linear flaw.

Further, in the rod flaw inspection apparatus and the rod flaw inspection method according to the present invention, when the number of pixels detected as flaws does not reach a preset reference number, the flaw detection density difference value (flaw detection) The flaw detection process can be repeated by changing the detection threshold value to a smaller value. Therefore,
When there is a deep and long linear flaw, it can be quickly determined to be defective, and a shallow linear flaw can be reliably detected. Further, the density difference value for flaw detection can be automatically set so that the number of pixels detected as flaws in the captured image of the entire peripheral surface of the rod reaches the reference number. Therefore, the grouping accuracy can be kept good, and the linear flaw can be detected with high accuracy.

Further, in the rod flaw inspection apparatus and the rod flaw inspection method according to the present invention, the density difference (luminance difference) between the pixels separated by two pixels in the rod axis direction is a preset density difference value for flaw detection (flaw detection). If the density difference exceeds the threshold value, the density difference gradient coefficient is set to 1. If the density difference between pixels separated by one pixel in the rod axis direction exceeds a preset density difference value for flaw detection, the density difference gradient coefficient is set to 1. Is set to 2, and the detection of the flaw pixel is weighted. If the cumulative value of the density difference coefficient does not reach the preset reference cumulative value, the flaw detection density difference value (flaw detection threshold value) is set to a smaller value. Since the detection of the flaw pixel is repeated by changing to the value, the density difference value for flaw detection can be automatically set so that the flaw portion can be accurately extracted. Since a portion where the density difference (luminance difference) between adjacent pixels is large is likely to be a flaw, when the density difference value for flaw detection between adjacent pixels exceeds a flaw detection threshold value, the weighting coefficient is increased. By that
It is possible to set a detection sensitivity at which a flaw can be accurately detected. Therefore, the detection sensitivity is set higher than necessary, and it is possible to prevent erroneous detection of a non-scratch noise portion as a scratch.

[0020]

Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a block diagram of a rod flaw inspection apparatus according to the present invention. A rod flaw inspection device 1 according to the present invention includes an imaging device 2 and an image processing device 3
And an image display device 4.

The imaging device 2 includes an imaging unit 21, an illumination light source 22, an imaging control unit 23, and a rotation mechanism unit 24. The imaging unit 21 is configured using a one-dimensional CCD camera. The imaging unit 21 is disposed to face the rod 5 with the axial direction of the rod 5 to be inspected as a visual field.
As the illumination light source 22, a linear light source is used so that uniform illumination light can be emitted in the axial direction of the rod 5. Rotating mechanism 2
4 holds the rod 5 mounted on a rod mounting unit (not shown), and rotates the rod 5 around its axis based on a rotation command signal 23a supplied from the imaging control unit 23.

When the image capturing command signal 37a is supplied from the image processing device 3, the image capturing control unit 23 controls the image capturing operation of the image capturing unit 21 and the rotating operation of the rotating mechanism unit 24, and The whole surface image is captured. The imaging control unit 23 supplies the imaging unit control information 23b to the imaging unit 21, thereby controlling the imaging timing of the imaging unit 21 and the output timing of the image signal (image signal for one line) 21a. The output timing signal 23c is supplied to the image processing device 3 prior to the output of the image signal (image signal for one line). When the imaging of one line and the output of the image signal 21a are completed, the imaging control unit 23
3a is supplied to the rotation mechanism 24 to rotate the rod 5 by a predetermined angle. The imaging control unit 23 captures a surface image over the entire circumference of the rod 5 by sequentially repeating imaging of the rod surface and rotation of the rod 5.

In the present embodiment, 3000
Using a one-dimensional CCD camera with pixels, an image is obtained in which the entire length of the rod 5 in the axial direction is 3000 pixels.
Further, the rod 5 is rotated by 1.8 degrees. As a result, the entire peripheral surface of the rod 5 becomes 3000 pixels ×
An image having 200 pixels is obtained. Note that the number of pixels of the captured image and the rotation angle of the rod 5 can be appropriately set according to the size of a flaw to be detected and the detection resolution. Further, in the present embodiment, a configuration in which a CCD camera having an imaging lens system is used for the imaging unit 21 has been described. However, a close contact type image sensor is arranged near the rod surface to obtain an image of the rod surface. You may. Furthermore, although the configuration in which the rod 5 is rotated has been described in the present embodiment, an image of the entire peripheral surface may be obtained by rotating the imaging unit 21 while maintaining a predetermined distance from the rod surface. . The rod 5 and the imaging unit 21 may be rotated in opposite directions. When the imaging unit 21 is rotated, the imaging unit 21 is rotated while maintaining the positional relationship between the imaging unit 21 and the illumination light source 22 in a fixed relationship. Alternatively, an image of the entire circumference of the rod may be obtained by continuously rotating the rod 5 at a predetermined speed and repeating the imaging at predetermined time intervals.

FIG. 2 is an explanatory diagram showing an example of the positional relationship between the illumination light source and the image pickup unit. In the present embodiment, the rod 5
The illumination light from the illumination light source 22 is irradiated at a predetermined incident angle θ1 with respect to the normal direction of the surface of
The imaging unit 21 is arranged in a direction (for example, a direction normal to the surface of the rod 5) different from the direction in which the angle θ2 and the emission angle θ2 are equal to each other to image the rod surface. Note that the imaging unit 21 may be arranged in the regular reflection direction. If there is a scratch on the rod surface, the reflection angle of the reflected light changes (due to the irregular reflection caused by the scratch), so that the amount of light received by the imaging unit 21 changes. This allows the captured image to change in shading (change in brightness)
Is generated, it is possible to detect a flaw based on a change in shading (change in brightness) of a captured image.

As shown in FIG. 1, the image processing unit 3
D converter 31 and image memory (frame memory) 32
A density difference value changing unit 33 for flaw detection, and a flaw pixel detection unit 34
A defect pixel grouping unit 35, a pass / fail determination unit 36, an image capture control unit 37, a pass / fail determination reference value storage unit 38, an operation unit 39, and a display control unit 40. Note that the image processing unit 3 can be constituted by a personal computer having an image input board and the like and an image processing program.

The A / D converter 31 A / D converts the image signal 21a output from the imaging unit 21 for each pixel, and converts the image signal on the rod surface into image data 31a. In this embodiment, a monochrome CCD camera is used for the imaging unit 21 and a gray-scale image data (luminance data) 31a of 256 gradations is obtained by using an A / D converter 31 having a resolution of 8 bits. I have to. In the present embodiment,
The configuration in which the image signal 21a output from the imaging unit 21 is directly input to the A / D converter 31 has been described.
An analog signal processing unit is provided in a stage preceding the analog signal processing unit 1, and the image signal 21 a is amplified and the DC level (average output level) is corrected by the analog signal processing unit.
An image signal is generated so that the dynamic range a (the minimum value and the maximum value of the voltage of the image signal) falls within the input voltage range of the A / D converter 31, and the generated image signal is sent to the A / D converter 31. You may make it input. With such a configuration, a minute change in the amount of reflected light due to a flaw or the like can be detected, and the detection resolution can be improved (the flaw detection capability can be improved). The image data 31a is supplied to the image memory 32,
Is temporarily stored.

When the inspection start request 39a is supplied from the operation section 39, the image capture control section 37 outputs the image capture command signal 3
7a is supplied to the imaging control unit 23 to start imaging the rod surface. The image capture control unit 37 controls the image capture timing based on the output timing signal 23c supplied from the imaging control unit 23. Image capture control unit 37
Converts the A / D conversion timing clock signal 37b into an A / D signal.
The image signal 21a is A / D-converted for each pixel by supplying it to the converter 31, and the write address information 37c is sequentially supplied to the image memory 32, so that the image data corresponding to each pixel is provided. 31a is written to a predetermined address. Thus, the image memory 32 stores a multi-valued (256 gradation) image composed of grayscale image data (luminance data) in which the axial direction of the rod 5 is the main scanning direction and the rotating direction of the rod is the sub-scanning direction. You.

FIG. 3 is an explanatory view of the rod surface image stored in the image memory. As shown in FIG. 3, the surface image of the entire circumference of the rod 5 has, for example, 3000 pixels in the axial direction (X direction) and 200 pixels in the circumferential direction (Y direction) of the rod 5 (2 pixels).
(00 line), each pixel is temporarily stored in the image memory 32 as image information composed of density data (luminance data) of 256 gradations.

The density difference value changing unit 33 for flaw detection shown in FIG.
Supplies a relatively large value (for example, 10) to the flaw pixel detection unit 34 as an initial value of the flaw detection density difference value 33a. The flaw pixel detection unit 34 reads the density data of one line in the axial direction from the image memory 32, temporarily stores the read density data in the density data storage unit in the flaw pixel detection unit 34, and reads the density data in the rod axis direction. Pixels whose density difference exceeds the preset density difference value for flaw detection are detected as flaws. Note that, without providing a density data storage unit for temporarily storing density data for one line in the flaw pixel detection unit 34,
The flaw pixel detection unit 34 may detect flaw pixels while sequentially reading density data from the image memory 32.

FIG. 4 is an explanatory diagram of the flaw detection processing in the flaw pixel detection unit. FIG. 4B shows an example of density data (density value) for one line in the axial direction. FIG. 4 (b)
In the figure, several pixels (11 pixels) in the axial direction are shown. FIG. 4A is shown as a graph for easy understanding of a change in density value (luminance value). FIG.
In (a), the vertical axis represents the density value (luminance value), and the horizontal axis represents the pixel number (corresponding to the position in the axial direction). Scratched pixel detection unit 34
Compares the density value of the pixel of interest with the immediately preceding pixel,
If the density increases by more than the density difference value for flaw detection (for example, 10), the pixel of interest is determined to be a flaw pixel, and the density of the previous pixel is higher than the density difference value for flaw detection (for example, 1).
0) In the case of the increase, the density difference gradient coefficient is set to 2. The flaw pixel detection unit 34 compares the density value of the pixel of interest with the pixel immediately before the pixel of interest, and if the density increases by a density difference value for flaw detection (for example, 10) or more, sets the pixel of interest to the flaw pixel. Is determined, and the density difference slope coefficient is set to 1 if the density has increased by more than the density difference value for flaw detection (for example, 10) with respect to the two pixels immediately before. FIG. 4C shows the detected density difference slope coefficient in association with the pixel position.

As described above, the flaw pixel detection unit 34 detects a flaw on the rod surface based on a change in which the density increases in the rod axis direction. The density change caused by the flaw includes a change in brightness and a change in darkness. Since these changes are paired as shown in FIG. 4A, the density (luminance) increases in this embodiment. The flaw detection is performed by paying attention only to the change that occurs. Note that the flaw may be detected based on a decrease in the density (luminance) instead of an increase in the density (luminance). Further, both an increase and a decrease in the density change (the luminance change) may be detected.

If the density increase change value from the immediately preceding pixel exceeds the density difference value for flaw detection, the flaw pixel detection unit 34 sets that pixel as a flaw pixel and sets the density difference gradient coefficient to 2, The coordinate data (Xn, Yn) of the flaw pixel and the density difference gradient coefficient are stored in the flaw detection result storage section in the flaw pixel detection section 34. If the density increase change value from the immediately preceding pixel does not exceed the density difference value for flaw detection, the flaw pixel detection unit 34
A determination is made as to whether the density increase change value from the immediately preceding pixel exceeds the density difference value for flaw detection, and if so, the density difference gradient coefficient is set to 1 and coordinate data (Xn, Y
n) and the density difference slope coefficient are stored in the flaw detection result storage section in the flaw pixel detection section 34. The flaw pixel detection unit 34 performs flaw detection processing in the axial direction for all pixels in the axial direction.
However, no processing is performed on the first pixel (X0) in the axial direction, and only the comparison with the previous pixel (X0) is performed on the second pixel (X1). The scratch pixel detection unit 34
A flaw detection process in the axial direction for each line is performed for all lines.

FIG. 5 is an explanatory diagram showing the result of flaw detection by the flaw pixel detection unit. FIG. 5A shows an example in which a flaw pixel detected with a density difference gradient coefficient of 2 and a flaw pixel detected with a density difference gradient coefficient of 1 coexist. FIG.
(B) shows an example in which only a flaw pixel detected as having a density difference slope coefficient of 2 exists. In FIG. 5, pixels detected as having a density difference gradient coefficient of 2 are shown in black, and pixels detected as having a density difference gradient coefficient of 1 are shown as hatched.

The flaw detection density difference value changing unit 33 shown in FIG.
Calculates the sum of the density difference gradient coefficients 34a for all the pixels detected as flaws by the flaw pixel detection unit 34.
When the sum of the density difference gradient coefficients is 1000 or less, the flaw detection density difference value changing unit 33 a
Is determined to be too high to detect the flaw on the rod surface properly, the flaw detection density difference value 33a is changed to a value lower than the initial value, and the flaw pixel detection unit 34 performs the flaw detection processing again. Let them do it. When the sum of the density difference gradient coefficients exceeds 1000, the flaw detection density difference value changing unit 33 transmits the flaw pixel grouping start request information 33b to the flaw pixel grouping unit 3.
5 to start flaw pixel grouping.

When the sum of the density difference gradient coefficients is more than 100 and not more than 1000, the flaw detection density difference value changing unit 33 lowers the flaw detection density difference value 33a by one level. That is, the density difference value 33a for flaw detection is changed from the initial value of 10 to 9
Change to When the sum of the density difference gradient coefficients is more than 10 and 100 or less,
The density difference value 33a for flaw detection is reduced by two levels. That is, the density difference value 33a for flaw detection is changed from the initial value of 10 to 8. When the sum of the density difference gradient coefficients is 10 or less, the flaw detection density difference value changing unit 33 lowers the flaw detection density difference value 33a by five levels. That is, the density difference value 33a for flaw detection is changed from the initial value of 10 to 5.

The flaw detection density difference value changing unit 33
The density difference value 33a for flaw detection newly set based on the sum of the density difference gradient coefficients of the flaw pixels detected by the previous flaw pixel detection processing is supplied to the flaw pixel detection unit 34, and is transmitted to the flaw pixel detection unit 34. A flaw pixel detection process based on the newly set flaw detection density difference value 33a is performed. If the sum of the density difference gradient coefficients exceeds 1000 as a result of the flaw pixel detection processing based on the newly set flaw detection density difference value 33a, the flaw detection density difference changing unit 33 Supply of the conversion start request information 33b to the damaged pixel grouping unit 35,
Start the wound pixel grouping. If the sum of the density difference gradient coefficients is 1000 or less as a result of the flaw pixel detection processing based on the newly set flaw detection density difference value 33a, The change of the density difference value is repeated until the sum of the density difference gradient coefficients exceeds 1000.

In the present embodiment, the processing for optimizing the density difference value for flaw detection based on the sum of the density difference gradient coefficients of the pixels detected as flaws has been described. The density difference value for flaw detection may be optimized so that the total number is equal to or more than the predetermined number of pixels.

The flaw pixel grouping section 35 receives the flaw pixel grouping start request information 33 from the flaw detection density difference value changing section 33.
When b is supplied, the coordinate data (Xn, Y) of the flaw pixel stored in the flaw detection result storage unit in the flaw pixel detection unit 34
n) Read all 34b and group the damaged pixels.

FIG. 6 is an explanatory diagram of the grouping process in the flaw pixel grouping unit. The scratch pixel grouping unit 35
First, after grouping in the axial direction (X direction), grouping is performed in the circumferential direction (Y direction). FIG. 6A shows a group range in the axial direction.
In the axial direction (X direction), other flaw pixels within ± 5 pixels in the axial direction (X direction) from the flaw pixel (pixel detected as flaw) are grouped.

After the grouping of the flawed pixels is completed for all the lines (all the axial directions), ± in the circumferential direction (the Y direction)
When another defective pixel or another defective pixel group exists within the range of 5 lines (pixels), the same if the displacement of each defective pixel or defective pixel group in the axial direction is within ± 1 pixel range. Group as a wound group. This circumferential grouping is performed over the entire circumference (all lines).

FIG. 6B shows a grouping range in the circumferential direction when the grouping in the axial direction is not performed (the number of flawed pixels = 1). In this case, the deviation of the axial position within ± 5 lines from the damaged pixel shown in black is ±
Damaged pixels in a range of one pixel, that is, a range indicated by hatching are grouped. FIG. 6C shows the grouping range in the circumferential direction with respect to the flaw groups grouped over 11 pixels in the axial direction. In this case, within a range of ± 5 lines from the damaged pixel group shown in black,
Flawed pixels whose axial position shifts by one pixel further in the axial direction from the grouped flaw range are grouped.

FIG. 7 is an explanatory diagram of another grouping process in the flaw pixel grouping unit. Wound pixel grouping unit 35
Indicates the distance from the scratch pixel (pixel detected as a scratch) in the axial direction (X
Other flawed pixels that are within ± 5 pixels in the (direction) and ± 5 pixels in the circumferential direction (Y direction) may be grouped. In FIGS. 6 and 7, the damaged pixel is shown in black regardless of the value of the density difference gradient coefficient. The range to be grouped is indicated by hatching.

The defective pixel grouping unit 35 sequentially groups defective pixels within a predetermined range by repeating the grouping process shown in FIG. 6 or FIG. When there are no more flawed pixels to be grouped around the newly grouped flawed pixels, the flawed pixel grouping unit 35 assigns a label name to the flawed pixel group obtained by a series of grouping, and performs grouping. The position coordinates of all the damaged pixels thus created are stored in the damaged pixel group storage unit in the damaged pixel grouping unit 35 in association with the label names. The damaged pixel grouping unit 35 performs a grouping process on other damaged pixels that have not been grouped. By this grouping process, a linear flaw is extracted.

When the grouping process is completed over the entire range of the rod surface, the flaw pixel grouping unit 35 sets the axial length (width) of the linear flaw and the circumferential direction of the linear flaw for each flaw group. Determine the length and area of the linear wound.

FIG. 8 is an explanatory diagram showing an example of a group of flawed pixels (linear flaws). FIG. 8 shows an example in which two linear flaws (label A and label B) are extracted by the grouping process. The flaw pixel grouping unit 35 calculates the difference between the maximum value of the axial coordinates (X coordinate) and the minimum value of the axial coordinates (X coordinate) from the position coordinates of each grouped pixel to obtain a linear flaw. The axial length (width) of the linear flaw is determined by calculating the difference between the maximum value of the circumferential coordinate (Y coordinate) and the minimum value of the circumferential coordinate (Y coordinate). Then, the total number of the scratched pixels in the group is defined as the area of the linear scratch.

The flaw pixel grouping unit 35 shown in FIG.
The data 35a of the width, length and area of the linear flaw are supplied to the pass / fail judgment unit 36 in association with the label name. The pass / fail determination unit 36 determines the size (width, length, area) of the linear flaw extracted by the flaw pixel grouping unit 35 and the pass / fail judgment criteria of the linear flaw stored in the pass / fail determination reference value storage unit 38. Value 38a
Is determined by comparing with. In the present embodiment, a linear flaw whose width, length, and area each exceed an allowable value (allowable width, allowable length, allowable area) is determined to be defective. It should be noted that any of the linear flaws whose width, length, or area exceeds an allowable value may be determined to be defective. In addition, even if the width, length, and area of the linear scratch do not exceed the allowable values (allowable width, allowable length, and allowable area), the total number of detected linear scratches exceeds the allowable number. May be determined to be defective.

The pass / fail judgment result 36a is supplied to the display control unit 40. The display control unit 40 causes the image display device 4 to display the pass / fail judgment result 36a. The image display device 4 is a CRT
It is configured using a display device, a liquid crystal display device, or the like. The display control unit 40 displays an operation menu on the screen of the image display device 4 to input a start request for image capture, set a pass / fail determination reference value, and perform various operations such as selection switching of a display image. Makes it possible to enter requests. When the pass / fail judgment reference value setting mode is designated, the display control unit 40 displays a menu screen for inputting the pass / fail judgment reference value, and prompts the user to input the pass / fail judgment reference value. The pass / fail judgment reference value 39b input from the operation unit 39 is stored in the pass / fail judgment reference value storage unit 38, and the input pass / fail judgment reference value 39b is displayed on the screen of the image display device 4. The pass / fail judgment reference value storage unit 38 is configured using a nonvolatile memory or a memory whose power supply is backed up by a battery or the like, and holds the input pass / fail judgment reference value 39b.

When a display request 39c for the rod surface image is supplied from the operation unit 39, the display control unit 40 generates a grayscale image of the rod surface based on the image data 32a stored in the image memory 32. Is displayed on the screen of the image display device 4. The display control unit 40 scrolls and displays the grayscale image of the rod surface based on the scroll request 39c from the operation unit 39. The display control unit 40 may display a pseudo color image instead of the grayscale image.

When the display control section 40 is supplied with the display request 39c of the defect pixel detection result from the operation section 39, the coordinate data of the defect pixel stored in the defect detection result storage section in the defect pixel detection section 34 ( Xn, Yn) and density difference slope coefficient 34c
, A pseudo color display image of the position of the flaw pixel and the density difference gradient coefficient is generated, and the detection result of the flaw pixel is displayed on the screen of the image display device 4 in a pseudo color.

When the display control section 40 is supplied with the display request 39c of the result of the processing for grouping the defective pixels from the operation section 39,
Based on the information 35b stored in the flaw pixel group storage section in the flaw pixel grouping section 35, a flaw pixel grouped image (linear flaw image) is generated and displayed on the screen of the image display device 4. . The display control unit 40 receives the defect determination result from the pass / fail determination unit 36 and the label name of the defective pixel group determined to be defective, and obtains a grouped image of the defective pixels determined to be defective (linear scratch image). ), Position information, and information such as the width, length, and area of the linear flaw may be displayed on the screen of the image display device 4 together with the result of the defect determination.

FIG. 9 is a flowchart showing the operation and processing of the rod flaw inspection apparatus and rod flaw inspection method. When an inspection start request 39a is input from the operation unit 39 in a state where the rod 5 to be inspected is mounted in the imaging device 2, the imaging by the imaging unit 21 and the rod rotation by the rotation mechanism unit 24 are synchronized. Then, a surface image is captured over the entire peripheral surface of the rod 5. Image signal 2 captured
1 a is A / D converted by the A / D converter 31 and then stored in the image memory 32. Thus, the input of the surface image of the entire peripheral surface of the rod is completed (step S1).

When the input of the front surface image (image capture) is completed, the amount of change in density difference (luminance difference) between pixels in the axial direction (X direction) is determined by the flaw pixel detection unit 34. Pixels exceeding the detection threshold are detected as flaws. When flaw detection processing in the axial direction (X direction) is performed for all axial directions (all lines) (Step S)
2) The sum of the density difference gradient coefficients of the pixels detected as flaws by the flaw detection density difference value changing unit 33 is obtained, or the total number of pixels detected as flaws is obtained (step S3). The density difference value changing unit 33 for flaw detection detects the sum of the density difference gradient coefficients of the pixels detected as flaws or the flaw detection when the total number of pixels detected as flaws does not reach a predetermined condition. The density difference value for use (threshold for flaw detection) is lowered (step S4). Scratched pixel detection unit 34
Is the changed density difference value for flaw detection (flaw detection threshold value)
, The flaw detection process in the axial direction (X direction) is performed again. (Step S2).

If it is confirmed in step S3 that the sum of the density difference gradient coefficients of the pixels detected as flaws or the total number of pixels detected as flaws has reached a predetermined condition, in step S5 The defective pixel grouping unit 35 groups defective pixels within a predetermined pixel interval in the axial direction (X direction). Next, in step S6, the flawed pixels or flawed pixel groups within the predetermined pixel interval in the circumferential direction (Y direction) are grouped. Linear flaws are extracted by the grouping in step S5 and step S6. The flaw pixel grouping unit 35 calculates the size (width, length, area) of the linear flaw extracted by the grouping (step S7). The pass / fail determination unit 36 determines the size (width, length, area) of the extracted linear scratch and a pass / fail determination reference value (allowable width,
(Permissible length, permissible area) to determine the quality (step S8).

[0054]

As described above, the rod flaw inspection apparatus and the rod flaw inspection method according to the present invention determine a flaw based on the density difference between pixels adjacent in the rod axis direction, and determine the flaw in the rod axis direction. Linear flaws are detected by grouping pixels whose intervals between pixels detected as flaws based on the density difference are within a preset pixel distance, and the width, length, Since the pass / fail judgment is performed based on the area, a linear scratch on the rod surface can be accurately detected and the pass / fail judgment can be made with high accuracy.

Since the flaw is determined based on the density difference between the pixels adjacent in the rod axis direction, even if the density of the detected image changes due to a change in the reflection coefficient of the rod surface or the intensity of the illumination light, the flaw is determined. Can be accurately detected. Further, the flaw can be accurately detected without correcting the darkening of the image around the visual field range of the imaging unit (so-called shading correction).

By grouping pixels whose intervals between pixels detected as flaws based on the density difference in the rod axis direction are within a predetermined pixel interval, a spiral linear flaw can be extracted. . And the width of the detected linear flaw,
The pass / fail judgment is made based on the length and the area. Therefore, even if there is a shallow point-like flaw on the rod surface, the point-like flaw and the linear flaw can be distinguished, and accurate pass / fail judgment can be made based on the linear flaw.

Further, in the rod flaw inspection apparatus and the rod flaw inspection method according to the present invention, when the number of pixels detected as flaws does not reach a preset reference number, the flaw detection density difference value is increased. By changing the value to a small value, the flaw detection process can be repeated. Therefore, when there is a deep and long linear flaw, it can be quickly determined to be defective, and also a shallow linear flaw can be reliably detected. Further, the density difference value for flaw detection can be automatically set so that the number of pixels detected as flaws in the captured image of the entire peripheral surface of the rod reaches the reference number. Therefore, the grouping accuracy can be kept good, and the linear flaw can be detected with high accuracy.

Further, in the rod flaw inspection apparatus and the rod flaw inspection method according to the present invention, when the density difference between pixels separated by two pixels in the rod axis direction exceeds a preset density difference value for flaw detection, Set the tilt coefficient to 1 and set 1 in the rod axis direction.
If the density difference between the pixels separated by a pixel exceeds a preset density difference value for flaw detection, the density difference gradient coefficient is set to 2 and the detection of the flaw pixel is weighted, and the cumulative value of the density difference coefficient is set in advance. If the reference cumulative value is not reached, the density difference value for flaw detection is changed to a smaller value and the detection of flaw pixels is repeated, so that the flaw detection density difference value is used so that flaws can be accurately extracted. Can be set automatically. Since there is a high probability that a portion where the density difference between adjacent pixels is large is a flaw, when the density difference value for flaw detection between adjacent pixels is exceeded, the weighting coefficient is increased to detect the flaw portion accurately. Can be set. Therefore, the detection sensitivity is set higher than necessary, and it is possible to prevent erroneous detection of a non-scratch noise portion as a scratch.

[Brief description of the drawings]

FIG. 1 is a block diagram of a rod damage inspection apparatus according to the present invention.

FIG. 2 is an explanatory diagram illustrating an example of a positional relationship between a light source for illumination and an imaging unit.

FIG. 3 is an explanatory diagram of a rod surface image stored in an image memory.

FIG. 4 is an explanatory diagram of a flaw detection process in a flaw pixel detection unit.

FIG. 5 is an explanatory diagram showing a result of flaw detection by the flaw pixel detection unit.

FIG. 6 is an explanatory diagram of a grouping process in a flaw pixel grouping unit.

FIG. 7 is an explanatory diagram of another grouping process in the flaw pixel grouping unit.

FIG. 8 is an explanatory diagram showing an example of a group of flawed pixels (linear flaws);

FIG. 9 is a flowchart showing the operation and processing of a rod flaw inspection apparatus and a rod flaw inspection method.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Rod flaw inspection apparatus, 2 ... Imaging apparatus, 3 ... Image processing apparatus, 4 ... Image display apparatus, 5 ... Rod (inspection object), 2
DESCRIPTION OF SYMBOLS 1 ... Image pick-up part (one-dimensional CCD camera), 22 ... Light source for illumination, 24 ... Rotating mechanism part, 31 ... A / D converter, 32 ... Image memory, 33 ... Density difference value change part for flaw detection, 34 ... Damaged pixel detection unit, 35: Damaged pixel grouping unit, 36: Pass / fail judgment unit, 37: Image capture control unit, 38: Pass / fail judgment reference value storage unit, 39: Operation unit, 40: Display control unit.

Claims (6)

[Claims] An imaging apparatus for imaging a rod surface over the entire circumference of a rod while relatively moving a rod to be inspected and an imaging unit for imaging the peripheral surface of the rod, and an image output from the imaging unit. A / D converter for converting an image signal of the rod surface into image data, an image memory for storing the A / D converted image data, and a density difference between pixels adjacent in the rod axis direction based on the image data A flaw pixel detection unit for detecting as a flaw a pixel which exceeds a preset density difference value for flaw detection, and a flaw pixel group for grouping pixels in which a distance between pixels detected as flaws is within a predetermined pixel interval. And a pass / fail determination unit that compares the width, length, and number of pixels of the group of flawed pixels with a predetermined pass / fail determination criterion to determine pass / fail. Inspection equipment. 2. A flaw detection density difference value changing unit for changing a flaw detection density difference value to a smaller value when the number of pixels detected as flaws does not reach a preset reference number. The rod damage inspection device according to claim 1, wherein: 3. The apparatus according to claim 2, wherein the flaw pixel detecting section is provided in the direction of the rod axis.
If the density difference between pixels separated by a pixel exceeds a preset density difference value for flaw detection, the density difference gradient coefficient is set to 1, and the density difference between pixels one pixel away in the rod axis direction is set to the preset flaw detection. If the density difference value exceeds the threshold value, the density difference slope coefficient is set to 2,
The flaw detection density difference value changing unit changes the flaw detection density difference value to a smaller value when the cumulative value of the density difference coefficient does not reach a preset reference cumulative value. 2. The rod damage inspection device according to 1. 4. An image of the rod surface is taken over the entire peripheral surface of the rod while relatively moving the rod to be inspected and an imaging unit for imaging the peripheral surface of the rod, and an image signal of the rod surface is converted into image data. After the pixel is detected as a flaw, the pixel whose density difference exceeds a preset flaw detection density difference value between pixels adjacent in the rod axis direction is detected as a flaw. Detecting linear flaws by grouping pixels whose intervals are within the preset pixel interval, and comparing the width, length, and area of the detected linear flaws with the preset pass / fail judgment criteria A rod flaw inspection method characterized by performing the following. 5. When the number of pixels detected as flaws does not reach a preset reference number, flaw detection sensitivity is changed by changing the flaw detection density difference value to a smaller value. The rod damage inspection method according to claim 4, characterized in that: 6. When the density difference between pixels separated by two pixels in the rod axis direction exceeds a preset density difference value for flaw detection, the density difference gradient coefficient is set to 1, and pixels separated by one pixel in the rod axis direction are set. If the density difference between the two exceeds the preset density difference value for flaw detection, the density difference slope coefficient is set to 2, and if the cumulative value of the density difference coefficient does not reach the preset reference cumulative value, The rod flaw inspection method according to claim 4, wherein the flaw detection sensitivity is changed by changing the density difference value to a smaller value.