JP3218908B2 - Surface flaw inspection method and apparatus - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a harmless linear non-uniformity which is continuously or discontinuously generated on a surface of a test object such as a steel plate and harmful linear marks, scratches and sporadic linear irregularities. TECHNICAL FIELD The present invention relates to a surface flaw inspection method and apparatus capable of identifying spots.

[0002]

2. Description of the Related Art Conventionally, as a surface flaw inspection apparatus for detecting flaws on a surface of a steel sheet or the like, an apparatus and a method using an optical system have been established as general techniques. An optical surface flaw inspection apparatus shown in FIG. 9 is well known. FIG.
No. 58, entitled "Method of Determining Surface Defects of a Strip". The surface flaw inspection apparatus shown in FIG. 9 includes a laser oscillator 10 as a light source, a rotating mirror 12 for scanning a laser beam from the laser oscillator 10 in the width direction of the steel plate (strip) S, and reflection on the surface of the steel plate S. A condensing lens 14 for condensing the focused laser light at one point, and a spatial filter (mask) 16 for applying the reflected laser light condensed by the condensing lens 14
And a photoelectric converter 18 that receives light via the.

The surface flaw inspection apparatus includes a control unit 20 for processing image data and an arithmetic processing unit 22 for arithmetically processing the output. The control unit 20 performs image processing on an output signal from the photoelectric converter 18, performs predetermined preprocessing, and is input to the arithmetic processing unit 22. In the arithmetic processing unit 22, the control unit 2
Feature parameters are extracted from the processing signal output from 0, the extracted feature parameters are compared with a preset flaw feature, flaws are identified, and flaw signals are edited.
The flaw signal after editing from the arithmetic processing unit 22 is transmitted to the typewriter 24.
Output to The result of the surface flaw inspection is printed and an expansion table is created.

In this surface flaw inspection apparatus, a laser beam is scanned while the steel sheet S is running in the direction of the arrow, and the reflected light at that time is received by the photoelectric converter 18, and the control unit 20 controls the profile method shown in FIG. The output signal from the photoelectric converter 18 is processed according to the following. The profile method will be described with reference to FIG. First, the traveling steel plate S is shown in FIG.
A predetermined size as shown in (A), for example, 250 to 5
The image is divided into image processing units 26 of 00 (mm × mm). Based on the image signal from the image processing unit 26, it corresponds to the image processing unit. For example, a digital map 26A is created as schematically shown in FIG. This pixel is usually selected to be about 2 to 5 mm due to restrictions on data processing time and the like.

The digital map 26A compares the output signal of each pixel based on the flaw in FIG. 10A with the preset levels r1 to r6 as shown in FIG.
The pixel is characterized based on the levels r1 to r6, such as the presence or absence of a flaw and the size of the flaw.

Next, a digital map 2 shown in FIG.
Based on 6A, the flaw signal of each pixel is integrated (projected) in the vertical direction to create a surface flaw profile by the histogram shown in FIG. The width W, length H and area S of the flaw are extracted from the flaw profile as parameters representing the characteristics of the surface flaw.

[0007] Further, through the same processing, for each image processing unit obtained by fractionating the entire steel sheet S, the number of flaws having a length H equal to or less than the reference and the number of point flaws N
A, the number of linear flaws NB from the number of flaws having a length H equal to or longer than the standard, the flaw occurrence position PP from the steel plate end, the polarity ratio VI of the signal level shown in FIG. 10C [Example: (r4 + r5 + r6) / Σ
ri], and various parameters such as sharpness HW (length H / width W) are extracted.

In the conventional surface flaw inspection apparatus, the arithmetic processing section 22 judges (identifies) the flaw type using the various parameters obtained as described above. The determination of the flaw type is performed, for example, according to a flowchart shown in FIG. In the figure, first, a sampling design such as determining a required number of samples according to the kind and degree of a flaw (if the frequency of occurrence is high, increase the number of samples) is performed (step S10). A new sample is collected (step S11). This sample is usually made on the basis of the size of the A4 size.

A characteristic pattern is visually extracted from the collected sample, a flaw characteristic map (map) is created for each flaw (step S12), and an image of a basic logic is created based on the flaw characteristic map (step S13). At the same time, a speckle (flaw) pattern is observed for each of the flaws in the sample (step S13), and a spatial filter that matches the speckle pattern is designed (step S1).
4, 15).

[0010] After the above preparation is completed, a test for creating judgment logic is performed on a large number of samples (step S16), and the characteristic parameters such as the width W and the length H described above are extracted. A feature amount such as whether to use the feature is determined (step S17). Thereafter, flaw type determination logic is created by combining the determined characteristic parameters or the like (step S19), and the logic is tested using a sample for confirmation, or is actually run online to perform the test (step S20). As a result of this test, if the judgment logic is inaccurate, the process returns to step S19, and the same operation is repeated to increase the accuracy of the judgment logic (step S2).
1).

As described in detail above, in the conventional inspection apparatus, the type of flaw is determined based on the above-described determination logic. The continuity flaws generated on the surface of the steel plate (strip) S shown in the conventional example include a flaw and a linear mark shown in FIG. And a linear pattern-like unevenness as shown in FIG. However, although this pattern-like linear unevenness is harmless, the pattern-like linear unevenness may be erroneously detected as a harmful flaw, and must be distinguished from other kinds of flaws.

[0012] The above-mentioned flaws are rubbed by a fixed object at a certain point on the process line and become flaws on a straight line. The effect of defective products on product cost and product supply is enormous. Therefore, it is very important to clearly distinguish the linear flaws such as the flaws, the linear marks and the dents from the continuity pattern and the linear unevenness.

[0013]

In the conventional method for determining a surface flaw, a surface flaw is determined based on flaw distribution form information in at least one of the longitudinal direction and the width direction of a strip such as a steel plate. As shown in FIG. 13, the distribution form information of the flaws is generated within the section by dividing the width direction into several portions 1W to 4W and the longitudinal direction for each unit length l ₁ (m). The flaw type information and the continuity flaw of the surface flaw are recognized based on the flaw type information obtained from the limited image area M of the belt-shaped body, and sporadic and continuity flaws are determined.

[0014] For this reason, thin pattern-like continuity flaws lack flaw type information in the limited image area if overlooked,
The distribution form information of the flaws in at least one of the longitudinal direction and the width direction of the strip cannot be accurately detected. For this reason, there is a possibility that sporadic detection is erroneously made. Also, if the detection level of flaw detection is set high so as to avoid overlooking, there is a disadvantage that thin pattern-like objects are recognized as defects and overdetection increases, and such surface flaw inspection equipment is used in production lines. There was a problem that it could not be done.

Further, the harmless linear non-uniformity which continuously occurs on the surface of the inspection object such as a steel plate and the harmful linear marks and flaws are distinguished by the type of flaw obtained from a limited image area. Since it is determined from the information and the distribution form information of the flaws in at least one of the longitudinal direction and the width direction of the strip, it is not possible to determine whether the flaws are continuous flaws occurring at the same width position.

As a result, it is not possible to discriminate harmless linear marks and flaws which are harmless, which are continuously generated on the surface of the inspection object such as a steel plate. Further, the lateral direction is divided into several parts, flaw type information obtained from a limited image area of the band occurring in the section, and flaws in at least one of the longitudinal direction and the width direction of the band. Since it is determined from the distribution form information, it is not possible to determine whether the defect is a continuity flaw occurring at the same width position.

For this reason, since the conventional continuity flaw does not indicate the exact same position in the width direction but indicates the range in which the width direction is divided into several parts, the judgment length and the judgment length are determined. It is necessary to take a sufficiently large number of flaws in the inside, and there is a disadvantage that the judgment accuracy is poor. In addition, since there is no logic to identify harmless linear unevenness and harmful linear marks or scratches that continuously occur on the surface of the inspection object such as a steel plate,
Although it is possible to determine the continuity flaw, it is not possible to determine the type of the flaw. Accordingly, the type of flaw generated on the steel plate or the like cannot be determined in this way, and therefore, the determination of the grade has further ambiguity.

The present invention has been made in view of the above-described problems, and provides a surface flaw inspection method and an apparatus for discriminating whether a flaw generated on a steel plate or the like is harmless or harmful. .

[0019]

As a means for solving the above-mentioned problems, a first aspect of the present invention recognizes the kind of flaw on the surface of a band based on image information obtained by optically scanning the surface of the band. A surface flaw inspection method, dividing the longitudinal direction of the band-shaped body surface into predetermined intervals, creating a flaw distribution in the longitudinal direction by a histogram method using image information obtained from each section,
The frequency distribution is calculated based on the flaw occurrence peak position in each flaw distribution.
Created and based on the standard deviation of the frequency distribution or the standard deviation.
The flaw occurrence peak in the current flaw judgment target is calculated using the
By examining the degree of separation between the
Ri, surface flaws inspection method characterized by the flaw determination target flaws species of the current to determine a constant. Further, in the first invention, before
The determination is based on the flaw occurrence peak position in the current flaw determination target.
And if the degree of separation from the frequency distribution is a predetermined amount or more,
A surface flaw inspection method characterized in that it is determined to be linear unevenness such as processing unevenness or dirt . Further, in the first invention, when it is determined that the image is not linear unevenness in the determination,
In this case, based on the feature amount obtained from the image information,
Surface flaw inspection method, wherein flaw types such as linear marks, scratches, and linear splinters are determined .

[0020] The second invention is a surface flaw inspection device which performs flaw species recognition of the band-like surface on the basis of image information obtained by scanning the strip surface optically, longitudinal of the strip surface Image obtained by dividing the direction into predetermined intervals and obtaining from each section
Flaw distribution in the longitudinal direction by histogram method according to the information
Flaw distribution creation means for creating flaws, and flaw occurrence in each flaw distribution
A frequency distribution generator that creates a frequency distribution based on peak positions
Steps and the standard deviation of the frequency distribution or the standard deviation
Flaw occurrence peak in the current flaw judgment target
By examining the degree of separation between the position and the frequency distribution,
And a determining means for determining the type of the current flaw determination target flaw . Also,
In a second aspect, the determining means includes a current flaw determination target.
Flaw occurrence peak position and the frequency distribution
In the case of a fixed amount or more, linear unevenness such as processing unevenness and dirt
A surface flaw inspection device characterized by determining that there is. According to a second aspect of the present invention, the determining means determines that a linear
If it is determined that the image information is not uneven, the image information
Based on the feature quantity obtained from, linear marks, scratches or
A surface flaw inspection apparatus characterized by judging a flaw type such as a linear barb .

[0021]

According to the present invention, there is provided a method for optically scanning the surface of a belt-like body.
When recognizing the type of flaw on the surface of the band based on the image information
The longitudinal direction of the surface of the strip is divided into predetermined intervals,
Histogram method based on image information obtained from the section
The flaw distribution in the longitudinal direction is created, and flaw generation in each flaw distribution is performed.
A frequency distribution is created based on the raw peak position, and the frequency distribution is calculated.
Using the standard deviation of
Flaw occurrence peak position and frequency distribution in flaw judgment target
By examining the degree of separation, the current flaw judgment
The flaw species of interest is a surface flaw inspection method and apparatus for determine a constant.

Further, there is provided a method and apparatus for sequentially updating the position information and the like of the flaw distribution in each section of the belt-like body surface and performing a surface flaw inspection based on the updated flaw distribution. And the occurrence of defects corresponding to harmless linear irregularities. Further, the type of the surface flaw of the band is determined from the flaw distribution, the flaw type information obtained from the limited image area, and the distribution form information of the defect in at least one of the longitudinal direction and the width direction of the band are used. A surface flaw inspection method and apparatus for determining whether a flaw is sporadic or continuity to identify a flaw type.

According to the present invention, information such as flaws and unevenness is continuously collected by a surface flaw inspection apparatus, flaw distribution (profile) is prepared at a predetermined interval by a histogram method, and the flaw type judgment result of the surface of the inspection object is obtained. Of these, the position information in the width direction of the test object and the profile peak position (Pin) obtained as the past performance profile using the position information of the tested result are used as reference positions to determine whether or not the defect is a continuously generated flaw. .

In this flaw type determination method, first, the current determined flaw type is (XX), and the flaw occurrence peak position of this determined flaw type is position information (PP). The threshold value n (number of flaws) is set by frequency distribution based on the PP values of all the judgment flaw types existing in the past Nm, and the actual profile for the past Nm is cut by the threshold value n. Standard deviation (σ), profile peak point position (PiP), median value (PiB), cut section width (PiW), profile peak value (Pin) with respect to the flaw distribution of the portion exceeding the threshold value n (hatched portion) And the like, and calculate data based on the statistical processing.

The PP value of the judgment flaw type (XX) is compared with the PiP value, and the PP value of the judgment flaw type (XX) is PiP.
Are compared for each peak position (PiP) in a certain set range (± Piw, Piw = PiW / 2). When the PP value of the judgment flaw type (XX) does not exist within a certain setting range (± Piw) of PiP, it is left to normal logic judgment.

When the PP value of the judgment flaw type (XX) is within a certain setting range (± Piw) of PiP, the PP value of the judgment flaw type (XX) is further set within a certain setting range of PiP (for example, ± Pw). nσ), and the PP value of the judgment flaw type (XX) is Pi
The section width range (for example, Piw) after cutting the actual profile peak value of P with respect to the past Nm at the threshold value n, and the set pixel range (for example, ± g, g) where the PP value of the judgment flaw type (XX) is PiP : The number of pixels), or sequentially narrowing the range to determine whether the continuity flaws have the same position and the same form.

As described above, by setting the numerical values set by the statistical processing in a predetermined setting range in advance, it is possible to set the optimum conditions according to the actual condition peculiar to the line. For example, the PP value of the judgment flaw type (XX) is within a set range of PiP (± P
iw), the feature quantity of the current determination flaw type is compared with the feature quantity profiles of all the determination flaw types existing in the past Nm to determine whether the flaws are continuous flaws of the same form. Make a decision.

If the flaws are not continuous flaws of the same form, the type of flaw is determined as sporadic flaws and is usually left to logic determination. If the continuity flaws have the same form, the PP value of the judgment flaw type (XX) is set within a certain range of PiP (eg, ±
nσ) The determination is made as a linear flaw, and the final determination is made through normal logic determination or linear flaw type determination dedicated logic. In this way, whether the flaws are continuity flaws of the same form is strictly evaluated, and when it is determined that the flaws are continuity flaws, the final determination is made through a logic dedicated to linear flaw type determination. The dedicated logic for determining the linear flaw type is a dedicated logic for determining the flaw type based on the current determined flaw type feature amount after determining as a linear flaw.

Further, the update and storage of the profile peak position (PiP) obtained as the past performance profile is as follows.
First, the determination length setting (N) of the band is set in advance.
The peak position (PiP) of the past performance profile is updated and stored according to the determination length setting value. In this way, the determination length setting (N) is set in advance, and even if different types of flaws or unevenness occur, the harmful linear unevenness and the harmful linear mark are accurately determined from the very short past length. And flaws such as flaws and sporadic linear barges can be identified.

[0030]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A surface flaw inspection method and a surface flaw inspection apparatus according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing one embodiment of a surface flaw inspection apparatus according to the present invention. In the figure, the surface flaw inspection device comprises an optical inspection device and an inspection device having a judgment processing function. The optical inspection apparatus irradiates the surface of the steel sheet S with the laser beam from the laser oscillator 10 via the rotary reflector 12, the reflected light is collected by the condenser lens 14, and the photoelectric converter through the spatial filter 16. The light is condensed at 18. The photoelectric converter 18 converts the diffraction light reflected from the object to be measured into an electric signal. Further, the inspection device having the judgment processing function includes a control unit 1 and an arithmetic processing unit 2. The control unit 1 includes an image processing unit 3 and a pre-processing unit 4. The arithmetic processing unit 2 includes a feature extracting unit 5, a flaw distribution creating unit 6 to be subjected to statistical processing, and an actual flaw distribution storage in which past actual profiles are written. It comprises means 7 and identification means 8.

In this surface flaw inspection apparatus, an output signal from the photoelectric converter 18 is input to the image processing unit 3, its output is input to the preprocessing unit 4, and the processed signal is input to the arithmetic processing unit 2. . The feature extraction means 5 of the arithmetic processing unit 2 performs image processing or the like on the processed signal to extract feature parameters, and the image signal is input to the flaw distribution creation means 6 and subjected to statistical processing.
A frequency distribution (histogram) as shown in FIG. 3 is created. Further, statistical processing is performed.

The flaw distribution data of the flaws generated on the steel sheet S in the past is written into the actual flaw distribution storage means 7, and the output data of the flaw distribution creating means 6 is the same as the data read from the flaw distribution storage means 7. Are compared to determine the type of flaw generated in the steel sheet S, the output result is displayed by the output stage 9 (CRT display or the like), and the output result is printed.

FIG. 5 is a diagram showing a typical flaw distribution (profile), in which the horizontal axis represents the width of the strip of steel (OP side to DR side).
Side), and the vertical axis indicates the number of flaws (pieces). In the drawing, P1 and P2 indicate peaks due to the flaw distribution.
n is a threshold value, PiP (i = 1, 2,...) is a peak point position, PiB is a median value, PiW is a width of a base (cut section width) of a peak of a hatched portion exceeding the threshold value n, and Pin is a peak value. I have. Further, Piσ (i = 1, 2,...) Is a standard deviation value of a hatched portion. Pi * (i = 1, 2,...) Is a simple average value of a hatched portion. Also, the peak point position P
iP indicates the position in the width direction of the steel sheet from the peak point of the peak Pi (i = 1, 2), and the reference is the distance from the edge or the center of the steel sheet. FIG. 5 shows a frequency distribution based on the PP values of all judgment flaw types existing in the past Nm, cuts the peak value of the performance profile (flaw distribution) for the past Nm at the threshold n, and then peaks the performance profile over the entire width. The position (PiP) is shown.

FIG. 4 shows a flaw distribution obtained by accumulating the flaw data of each section in the longitudinal direction at predetermined intervals in the longitudinal direction of the steel sheet. It moves and a flaw distribution is created by the histogram method at a short period Nm (m = 1, 2,...). The flaw type is determined from the flaw distribution. Describing the flaw type with reference to FIG. 4, the line bar is displayed as the flaw distribution Pa in the cycle N2. The linear mark (unevenness) and the roving are displayed as the flaw distribution Pb in the cycle N3. Foreign matter is flaw distribution Pc in cycle N4
Will be displayed as The flaws that occur continuously with the periods N4 and N5 are displayed as the flaw distribution Pd in the periods N5, N6,. The sliver is displayed as a flaw distribution Pe in the cycle N6. The gouge is displayed as a flaw distribution Pf in the cycle N6.
The line fray (scratch flaw) is displayed as a flaw distribution Pf in the cycle N7. As described above, the characteristics of the flaws generated in the strip can be analyzed by the flaw distribution (profile).

FIG. 6 shows an image signal indicating the distribution of the flaws in one section and the output result thereof. The flaw distributions P1 and P2 are created by integrating dot-like flaws in the longitudinal direction. A flaw distribution has been created. Usually, this integration process is performed not by hardware but by software by a computer.

Next, the flaw type determining means in the surface flaw inspection apparatus of the present invention will be described with reference to the flowchart of FIG. In FIG. 5, in step S1, a determination result (XX) of a flaw distribution (profile) as shown in FIG. In step S2, frequency distribution is performed based on the PP values of all the judgment flaw types existing in the past Nm, and the peak value of the actual profile (flaw distribution) for the past Nm is cut by the threshold n,
Peak position (Pi) of flaw distribution (actual profile) over the entire width
P).

In step S3, it is determined whether or not the current PP value of the logic determination flaw type (XX) is within the range (± Piw) of the cut section width after cutting at the PiP threshold value n. If it is within ± Piw), the process proceeds to the next step S4. If it is outside the range, it is recognized as a single sporadic flaw, and the current logic judgment flaw type (XX) is output as a result.

In step S4, the PP value of the current logic judgment flaw type (XX) is within a set range of PiP (for example, P
iP ± 3σ) to determine whether or not it exists within the range (Pi
If it is within (P ± 3σ), the process proceeds to the next step S5. If it is outside the range, it is recognized as a single sporadic flaw, and the current logic judgment flaw type (XX) or the linear shape of the pattern in FIG. The final decision is made through the logic dedicated to flaw type decision.

In step S5, it is determined whether or not the PP value of the current logic determination flaw type (XX) is within a certain set range of PiP (for example, PiP ± 1σ). If it is, the process proceeds to the next step S6. If it is out of the range, it is recognized as a single sporadic flaw, and the current logic judgment flaw type (XX) or the linear flaw type judgment of the pattern of FIG. Final decision is made through the logic of

In step S6, the type of linear flaw in the pattern of FIG. 3A is determined, and then the type of linear flaw is determined.
If it is the flaw type existing in the past Nm and the feature amount of the judgment flaw type existing in the PiP set range (PiP ± g), the process proceeds to step S7.
Go to 9. In step S7, it is determined whether or not the width is small. In step S9, it is determined whether or not the feature amount distribution of the past Nm is the same occurrence position or the same occurrence form. If it is determined in step S9 that they are the same, the process proceeds to step S10. In step S10, the occurrence ratio of the intermediate width is determined.

Further, in step S8, the type of the flaw is determined based on the occurrence rate of the polarity of the positive polarity or the negative polarity in step S8. By looking at the feature amount of the flaw type more closely in this way, from a very short past length, accurate harmless linear irregularities and harmful linear marks, scratches and sporadic linear scabs are obtained. It becomes possible to identify.

FIG. 3B shows a linear flaw type classified into several patterns. The pattern is a linear flaw type (for example, at the same position in the width direction and linear in the length direction). Linear marks, linear irregularities, roping, scratches, etc.) and patterns are concentrated at substantially the same position within a certain range in the width direction, and are continuously generated in the length direction (for example, wavy marks, button marks). The pattern is within a considerably wide range in the width direction, and furthermore, it is conceivable that there are flaw types (for example, processing unevenness, dirt, etc.) which are continuously generated in the length direction, etc. It is important to distinguish between certain linear irregularities and harmful linear marks, scratches and sporadic linear scabs.

Next, another embodiment of the surface flaw inspection apparatus and the surface flaw inspection method according to the present invention will be described. FIG.
FIG. 4 is a block diagram showing another embodiment of the surface flaw inspection apparatus according to the present invention. 4, the flaw distribution data of the flaw distribution creating means 6 based on the statistical processing of the arithmetic processing unit 2 stores the flaw distribution data for each statistical section Nm (m = 1, 2,...) In the storage means 11 as shown in FIG. Is read. The representative actual flaw distribution is selected by the flaw type selection means 13 and stored in the actual flaw distribution storage means 7. The flaw distribution data read from the actual flaw distribution storage means 7 is compared with the current flaw distribution data by the identification means 8 to identify the flaw type.

The flaw type selection means 13 is performed according to the flowchart of FIG. In FIG. 8, in step S1, a flaw distribution is created, the flaw type is determined by the statistical processing described above, and the process proceeds to step S2, where the grade is determined. Subsequently, the process proceeds to step S3, where flaw distribution data is output. In step S4, the accuracy data of the digitized flaw distribution data is collected. In step S5, data collection conditions are set, and in step S6, the flaw distribution data set in the conditions of step S5 is stored. Statistical processing is performed based on the number of flaws in the statistical section Nm, and the data of the flaw type of each statistical section Nm is transferred to and stored in the storage means 11 such as a magneto-optical disk, and the data stored in the storage means 11 is The flaw type is selected by the flaw type selection means 13 and input to the actual flaw distribution storage means 7. Since this data is updated in a maximum of 2 seconds, it is possible to determine the type of a newly generated flaw in real time.

As described above, the surface flaw inspection method and the surface flaw inspection apparatus of the present invention use the flaw distribution created by the histogram method to determine the peak point position P according to the flaw type.
iP, median value PiB, cut section width PiW, peak value Pi
Since n, standard deviation value Piσ, and simple average value Pi * are different, the flaw type can be determined by detecting and comparing these values. These statistical processes are created by arithmetic processing by a computer. According to the present invention, harmless linear unevenness generated on the steel sheet S and harmful linear marks, scratches, and sporadic linear barges can be identified.

[0046]

As described above, the present invention relates to a method and apparatus for converting the diffraction light reflected from an object to be measured into an electric signal by using an optical system and inspecting a surface flaw of a band such as a steel plate. Yes, harmless linear unevenness, harmful linear marks, and flaws by referring to the position peak information obtained from the position information of the steel plate and the position of the inspected test object as the past performance profile. And sporadic linear barbs. Further, by updating and storing data of a new flaw distribution (profile), there is an advantage that a flaw which may occur in the future can be easily determined.

Therefore, harmless linear marks and harmful linear marks, scratches and sporadic linear scabs, which have been considered difficult in the past, can be distinguished from each other. In addition, the operation and the yield can be improved, and the production of high quality products can be achieved.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a surface flaw inspection method and a surface flaw inspection apparatus according to the present invention.

FIG. 2 is a flowchart for identifying harmless linear unevenness and harmfulness in FIG. 1;

FIGS. 3 (a) to 3 (c) are diagrams showing the occurrence patterns of linear unevenness and harmful linear marks, scratches, and sporadic linear barbs.

FIG. 4 is a diagram showing a flaw distribution in a predetermined cycle and a continuous distribution state.

FIG. 5 is a diagram showing a typical flaw distribution state.

FIG. 6 is a diagram showing a distribution of flaws obtained by image processing and a flaw distribution state obtained by data processing.

FIG. 7 is a block diagram showing one embodiment of a surface flaw inspection method and a surface flaw inspection apparatus according to the present invention.

FIG. 8 is a diagram showing a flowchart for creating flaw distribution data to be stored.

FIG. 9 is a block diagram showing an example of a conventional surface flaw inspection device.

FIG. 10 is a diagram showing an example of signal processing in a conventional surface flaw inspection device.

FIG. 11 is a flowchart showing a procedure for creating a conventional flaw type determination logic.

FIG. 12 is an explanatory diagram showing a flaw and a flaw.

FIG. 13 is a view for explaining a conventional method for determining flaws and unevenness occurring in a steel sheet.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Control part 2 Operation processing part 3 Image processing part 4 Preprocessing part 5 Feature extraction means 6 Flaw distribution creation means 7 Actual flaw distribution storage means 8 Identification means 9 Output means 10 Laser oscillator 12 Rotating mirror 13 Flaw type selection means 14 Light collection Lens 16 Spatial filter 18 Photoelectric converter

──────────────────────────────────────────────────続 き Continuation of the front page (72) Inventor Teruo Horisawa 1-2-1 Marunouchi, Chiyoda-ku, Tokyo Inside Nippon Kokan Co., Ltd. (56) References JP-A-2-90046 (JP, A) JP-A Sho-59 -99238 (JP, A) JP-A-7-49314 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G01N 21/84-21/958

Claims (6)

(57) [Claims] 1. A surface flaw inspection method for recognizing a flaw type of a surface of a band based on image information obtained by optically scanning the surface of the band, wherein the longitudinal direction of the surface of the band is divided into predetermined intervals. The flaw distribution in the longitudinal direction is created by the histogram method based on the image information obtained from each section, and the flaw occurrence peaks in each flaw distribution are generated.
Create a frequency distribution based on the network position and mark the frequency distribution.
Using the standard deviation or a value based on the standard deviation,
Between the flaw occurrence peak position in the judgment object and the frequency distribution
By examining the degree of separation, the current flaw judgment target
Surface defect inspection method, characterized by determine constant flaws species. 2. The method according to claim 1, wherein the determination is performed on a current flaw determination target.
The flaw occurrence peak position and the frequency distribution are separated by a predetermined amount
In the above case, it is linear unevenness such as processing unevenness or dirt.
2. The surface flaw inspection method according to claim 1 , wherein the determination is made as follows. 3. The method according to claim 1, wherein the determination is not linear unevenness.
If so, further features obtained from the image information
Based on the amount, type of flaws such as linear marks, flaws or linear scabs
Surface defect inspection method according to claim 2, wherein the determining. 4. A surface flaw inspection apparatus for recognizing a flaw type on a surface of a band based on image information obtained by optically scanning the surface of the band, wherein the longitudinal direction of the surface of the band is divided into predetermined intervals, From each section
According to the obtained image information,
Flaw distribution creation means for creating flaw distribution in the direction,
Each time a frequency distribution is created based on the peak position of the flaw occurrence
Number distribution creating means, the standard deviation of the frequency distribution or a value based on the standard deviation
By using the flaw occurrence peak position in the current flaw determination target and
By examining the degree of separation from the frequency distribution,
A surface flaw inspection apparatus, comprising: a determination means for determining a type of a flaw to be currently determined for flaw determination . 5. The method according to claim 1, wherein the determining unit determines a current flaw determination target.
Flaw occurrence peak position and the frequency distribution
In the case of a fixed amount or more, linear unevenness such as processing unevenness and dirt
The surface flaw inspection device according to claim 4, wherein it is determined that there is a surface flaw. 6. When the determination means determines that the image is not linear unevenness as a result of the determination, the determination means further determines a linear mark, a flaw, a linear shave, or the like based on a feature amount obtained from the image information. The surface flaw inspection apparatus according to claim 5 , wherein the flaw type is determined.