JP2005265639A - Method and device of detecting roll scratch in strip body - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method and a device of detecting a roll scratch in a strip body, which enables accurate judgment even when somewhat periodical fluctuation or noises are included. <P>SOLUTION: A method of detecting a roll scratch in a strip body, which comprises photographing the surface of a moving strip body, image-processing the image photographed and then detecting a roll scratch, the method comprising extracting a plurality of candidates for scratches appearing at a interval in the longitudinal direction of the strip body from the image photographed, creating a signal waveform having information about the number of the candidates for scratches and the longitudinal interval, and having an intended periodical variation width of the roll scratch from coordinates of the extracted candidates for scratches, calculating a self correlation function of the signal waveform, detecting a peak obtained from the calculation result, and judging the presence or absence of periodic property of the signal waveform, thereby detecting the roll scratch. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a belt-like roll wrinkle detection method and apparatus for optically detecting surface wrinkles, particularly periodically generated roll wrinkles, while moving a band made of metal, plastic or other material.

In order to improve the quality of products, surface defect inspection for optically inspecting the surface of a moving strip is widely performed.
In the surface wrinkle inspection, the surface is imaged with an imaging device while feeding the belt in the longitudinal direction thereof, and wrinkle candidates are extracted from the captured image by image processing. Finally, the position and size of the wrinkle candidate are determined. In addition, it is determined whether the wrinkle candidate is harmful or harmless based on a feature quantity such as shape, brightness, periodicity, or an image.
For example, in a rolling line such as a steel plate, oil, water droplets, light stains, and the like often adhere to the surface of the steel plate, and these oils and the like should be judged harmless unless they affect the quality of the final product.
On the other hand, a roll wrinkle, that is, a roll wrinkle generated by chipping or dents on the surface of the rolling roll or adhering foreign matter, which is transferred to the material to be rolled, is rolled each time the rolling roll rotates. Since it continues to occur in the longitudinal direction of the material at a period corresponding to the roll circumference, it should be judged harmful. Usually, when this roll wrinkle or its periodicity is detected, an alarm is notified to the operator, and the work of exchanging the rolling roll that causes the wrinkle is performed.
As described above, various wrinkles and dirt adhere to the surface of the steel plate moving on the rolling line, so how accurately and how quickly the roll wrinkle or its periodicity is detected from the wrinkle candidates. This is an important factor for improving the reliability of surface flaw inspection and thus the quality of the product.

Conventionally, many methods for detecting the roll wrinkle have been proposed, focusing on the periodicity of the roll wrinkles.
For example, the simplest method utilizing this periodicity is that roll rolls are generated at a period corresponding to the roll circumference in the longitudinal direction of the material to be rolled. If they match, it is determined that there is a periodicity and roll wrinkles are occurring.
However, in the rolling process of an actual steel sheet or the like, when wrinkles generated in the rolling roll are always transferred to the material to be rolled at a certain level due to fluctuations in the rolling reduction due to thickness variation of the material to be rolled. Is not limited. That is, in general, when the rolling reduction is small, roll wrinkles that cause fatal damage to the quality of the product do not occur, and when they are generated, if the level to be transferred is small, they are not extracted as wrinkle candidates.
For this reason, a so-called extraction omission occurs where the wrinkle candidates are extracted thinly, so that the roll circumference does not match the distance between the wrinkle candidates, and the roll wrinkle actually occurs, There was a problem that the roll wrinkle or its periodicity could not be detected.

As a method for solving the above problem, a detection method using an autocorrelation function is known (see, for example, Patent Document 1).
The autocorrelation function is widely used in the field of signal processing, particularly in the field of image processing, and the distance between peaks obtained from the calculation result of the autocorrelation function represents the period of the periodic signal component. Without knowing the period of the periodic signal component included in the signal sequence, only the periodic signal component can be selectively extracted from the signal sequence buried in noise.
In other words, the method for detecting roll wrinkles using an autocorrelation function is to first calculate the autocorrelation function for the longitudinal data series of the wrinkle candidates and then use the peak obtained by this calculation. In this method, it is determined whether or not the data series has periodicity, and if there is periodicity, this is determined as a roll cage.
Therefore, according to this method, it is possible to detect a roll wrinkle having an arbitrary cycle without confirming the roll wrinkle cycle in advance, and even if the above-described extraction failure of the wrinkle candidate occurs.

However, in the actual rolling process of a steel plate or the like, the distance in the longitudinal direction of the roll ridges varies depending on the difference in rolls due to creases, the difference in rolling reduction, or the roll slip between the roll and the material to be rolled. That is, in general, a roll wrinkle has periodicity, but its generation cycle can vary due to various factors.
Further, as described above, the surface state of the steel sheet is various, and there are many noises due to surface unevenness, rough skin, etc., and there are also troublesome noises that occur periodically.
For this reason, even if an autocorrelation function that is effective for extracting periodic signal components is used, a periodically rolling roll ridge cannot be detected, and a periodically generated noise is erroneously determined as a roll ridge. there were.
Japanese Patent Laid-Open No. 2-74852 (2nd page)

  The problem to be solved by the present invention is to provide a method and apparatus for detecting roll wrinkles of a belt-like body that can be determined with high accuracy even if some periodic fluctuations and noise are included.

  In order to solve the above-mentioned problems, the present inventor has conducted a number of theoretical and experimental studies on a method and apparatus for detecting roll wrinkles in a strip, and as a result, has imaged the surface of the strip and spaced the strip in the longitudinal direction. A plurality of wrinkle candidates appearing at a position are extracted by image processing of the captured image, and a signal waveform is created from the extracted coordinates of the wrinkle candidate using the longitudinal coordinate function devised by the present inventor. The principle that the periodicity of the roll wrinkles can be accurately grasped by detecting the peak obtained from the result of the calculation and determining the presence or absence of the periodicity of the signal waveform based on this peak was found. .

First, the principle will be described with reference to FIG.
FIG. 1 (d) is a plot of the extracted wrinkle candidates, and assumes a case in which extraction is lost at the third position from the right and periodic fluctuation occurs at the fifth position from the right. In this case, the noise component is not considered.

FIG. 1 (a) is a diagram showing that the periodicity of a wrinkle candidate can be detected by using the autocorrelation function even when there is omission of the haze candidate extraction, and is a premise of the principle It is.
A1 in FIG. 1A is a rectangular display using the extracted coordinates of the wrinkle candidate as a rectangle with a part with a wrinkle candidate being 1 and a part without a wrinkle candidate being 0, and A2 and A3 are phase-shifted from A1, respectively. Is displayed by shifting to the left.
A4 indicates the number of candidates, A5 indicates the sum of the products of A1 and A2, and A6 indicates the sum of the products of A1 and A3. These are the peaks obtained as a result of the calculation of the autocorrelation function. Correspondingly, the distance between the peaks indicates the length of the period.
That is, this indicates that the periodicity of the wrinkle candidate can be detected by using the autocorrelation function even when the wrinkle candidate is missing.

Next, FIG. 1B shows a case where a signal waveform is created by the same method as the above method in the case where a period variation occurs in addition to the omission of the defect candidate extraction.
In this case, the levels of the detected peaks B5 and B6 are lower than the values of A5 and A6 in FIG. 1A, and the peak cannot be detected depending on the set detection threshold.
This indicates that when periodic fluctuations occur in the roll ridge, it is difficult to reliably grasp the periodicity because the level of the peak indicating the period is low even if the autocorrelation function is simply used. .

Based on these, the present inventor has devised a signal waveform using a longitudinal coordinate function indicated by C1 in FIG.
C1 in FIG. 1C is created by giving a width of an assumed periodic variation ΔY to the rectangle shown by A1 in FIG. 1A and B1 in FIG. 1B.
That is, C1 in FIG. 1 (c) is a rectangle displayed by using the extracted coordinates of the candidate for wrinkles as a rectangle with a portion having a wrinkle candidate as 1 and a portion without a wrinkle candidate as 0, but the width of the rectangle is assumed. The width of the periodic fluctuation ΔY is created.
In other words, C1 is a longitudinal coordinate function which is a signal waveform represented by a rectangular wave having information on the number of wrinkle candidates and the distance in the longitudinal direction and having an assumed periodic variation width of the roll wrinkles.
C2 and C3 are displayed with the phase shifted from C1 to the left.
C4 indicates the number of candidates, C5 indicates the sum of products of C1 and C2, and C6 indicates the sum of products of C1 and C3. These are the peaks obtained as a result of the calculation of the autocorrelation function. Correspondingly, the distance between the peaks indicates the length of the period.
The peak value at this time is higher than that in FIG. 1B, and it is confirmed that the peak value has reached the same level as the peak value in FIG. be able to.
In other words, even if omission candidate extraction or period variation occurs, a signal waveform having the above characteristics is created from the extracted coordinates of the omission candidate, an autocorrelation function of this signal waveform is calculated, and this calculation is performed. If the peak obtained from the result is detected and the presence / absence of periodicity of the signal waveform is determined based on the peak, it means that the periodicity of the roll can be accurately detected.

  The above is the explanation of the principle, but the steel plate surface moving on the rolling line may have a lot of noise due to surface unevenness, rough skin, etc., and there are also troublesome noises that occur periodically. As is well known, it acts in the direction of lowering.

In order to solve the above problems, the present inventor has conducted a number of theoretical and experimental studies on the method and apparatus for detecting roll wrinkles in a strip, and as a result, there is a high similarity between wrinkles generated by transcription. I found.
That is, the roll ridge is a crease that is generated when chips or dents are formed on the surface of the rolling roll or foreign matter adheres to it, which is transferred to the material to be rolled, and each time the rolling roll rotates, Since the wrinkles continue to occur in the longitudinal direction of the material at a period corresponding to the roll peripheral length, the roll wrinkles have high similarity in shape, brightness, area, and the like.
Similarly, harmless soot generated by transferring oil, water droplets and the like attached to the rolling roll to the material to be rolled has high similarity in shape, brightness, area, and the like.
Therefore, the present inventor has found that the periodicity of the roll wrinkles can be accurately detected by the above method by using these similarities and removing the noise in advance.
Moreover, when there was little noise, it discovered that the periodicity of a roll wrinkle could be detected correctly by the said method, without removing the said noise beforehand.

  Based on the above knowledge, the present inventor has come up with a method and apparatus for detecting a roll wrinkle of a strip that can be determined with high accuracy even if some periodic fluctuations and noise are included. The gist is as follows.

(1) In a roll-shaped roll wrinkle detection method for detecting a roll wrinkle by imaging the surface of a moving belt-shaped body and processing the captured image to detect a roll wrinkle, a plurality of wrinkle candidates appearing at intervals in the longitudinal direction of the band-shaped body Extracted from the captured image, and from the extracted coordinates of the wrinkle candidate, creates a signal waveform having information on the number of wrinkle candidates and the distance in the longitudinal direction and having the expected fluctuation width of the roll wrinkle. An autocorrelation function is calculated, a peak obtained from the result of the calculation is detected, and roll wrinkles are detected by determining the presence or absence of periodicity of the signal waveform based on the peak. Roll wrinkle detection method.

(2) In a roll-shaped roll wrinkle detection method in which the surface of a moving belt-like body is imaged and the picked-up image is processed to detect roll wrinkles, a plurality of wrinkle candidates appearing at intervals in the longitudinal direction of the band-shaped body The image is extracted from the captured image, noise is removed from the extracted candy candidates, and the candy candidates are selected. From the selected coordinates of the candy candidates, the number of candy candidates and information on the longitudinal interval are included, and an assumed roll 疵A signal waveform having a period fluctuation width of λ is created, an autocorrelation function of the signal waveform is calculated, a peak obtained from the calculation result is detected, and the presence or absence of periodicity of the signal waveform is determined based on the peak Thus, a roll wrinkle detection method for a band-shaped body, characterized in that a roll wrinkle is detected.

(3) The method for detecting roll wrinkles of a band according to (2), wherein an identification method using a feature amount is used for selecting the wrinkle candidates.

(4) The method for detecting roll wrinkles of a band according to (2), wherein a pattern matching method using an image is used for selecting the wrinkle candidates.

(5) The strip-shaped roll wrinkle detection method according to (2), wherein both the identification method using a feature amount and the pattern matching method using an image are used for selecting the wrinkle candidate.

(6) Any one of (1) to (5), wherein the peak detection selects a peak whose height exceeds a preset threshold and has a minimum period among the assumed periods. 2. A method for detecting roll wrinkles of a band according to the above.

(7) A series of processing from creation of the signal waveform to determination of presence / absence of periodicity of the signal waveform is performed for each processing region divided into a plurality in the plate width direction of the strip (1) ) To (6) The method for detecting roll wrinkles of a band according to any one of (6).

(8) The method for detecting roll wrinkles of a band according to (7), wherein the plurality of processing regions are divided with a lapping margin in a plate width direction of the band.

(9) A signal used as an input signal of an autocorrelation function in a roll wrinkle detection method using an autocorrelation function that detects a roll wrinkle by imaging the surface of a moving band and performing image processing on the captured image. In the waveform creation method, using the coordinates of the wrinkle candidate extracted from the captured image, a part with a wrinkle candidate is created with a rectangular wave with a height of 1 and a part without a wrinkle candidate with a height of 0, and the width of the rectangle is assumed. A method for creating a signal waveform, characterized in that the period fluctuation width of the roll roll is set.

(10) In a roll-shaped roll wrinkle detecting device that picks up an image of the surface of a moving belt-shaped body and detects a roll wrinkle by performing image processing on the captured image, an imaging device that images the surface of the band-shaped body, and a longitudinal direction of the band-shaped body A means for extracting a plurality of wrinkle candidates appearing at intervals from the captured image, and information on the number of wrinkle candidates and the longitudinal interval from the extracted wrinkle candidate coordinates, and an assumed periodic variation of the roll wrinkles Means for generating a signal waveform having a width; means for calculating an autocorrelation function of the signal waveform; means for detecting a peak obtained from a result of the calculation; and presence or absence of periodicity of the signal waveform based on the peak An apparatus for detecting roll wrinkles of a belt-like body, comprising: an image processing apparatus having means for determining

  According to the present invention, the periodicity can be detected correctly and the reliability of the surface flaw inspection can be improved even if there is a slight fluctuation in the roll wrinkles due to roll slip, fluctuations in sheet thickness after rolling, and the like.

Hereinafter, the best mode for carrying out the present invention will be described with reference to FIGS. 2 is a schematic diagram of a surface wrinkle inspection apparatus, FIG. 3 is a block diagram of the main part of the apparatus, FIG. 4 is a flowchart showing one form of image processing in the present invention, and FIG. FIG. 6 is a map plotting the extracted wrinkle candidates.
Hereinafter, the case where a strip | belt body is a strip | belt-shaped steel plate is demonstrated.

The surface of the strip-shaped steel plate 1 moving in the plate passing direction is photographed by the imaging device 4.
As the imaging device 4, there are an area camera (sensor) method for reading an image in units of frames and a line camera (sensor) method for reading an image in units of lines, both of which can be used. Is fast, and when a high speed image processing is required, a line camera (sensor) system such as a line scan camera is desirable.
Moreover, as a picked-up image, there are a monochrome grayscale image and a color image, and both can be used.
Hereinafter, a case where a line camera that outputs a monochrome grayscale image is used will be described.
The imaging device 4 captures a grayscale image in the plate width direction (X direction) of the strip-shaped steel plate 1 and outputs an information amount of black and white 256 gradations.
The captured image is transferred to the main storage device 11 via the data input / output control device 12 to a computer that is the image processing device 9.
The data input / output control device 12 controls the imaging device 4 in synchronization with the rolling speed so that the imaging device 4 can image the surface of the strip steel plate 1 moving in the longitudinal direction (Y direction) without gaps. Output timing.
The imaging device 4 repeats imaging and transfer of the captured image at the control timing output from the data input / output control device 12.

The image processing program is stored in an auxiliary storage device such as a hard disk.
Setting items such as a threshold necessary for image processing are input from the input device 5 such as a keyboard and stored in the auxiliary storage device 8.
The captured image data subjected to the image processing is stored in the auxiliary storage device 8 and is output to the display device 7 via the graphics board 6.
The display device 7 displays harmful / non-hazardous determination and periodicity determination result together with the image of the steel plate surface.
The image processing will be described with reference to the flowchart of FIG.

Step S11
A black and white grayscale image captured by the imaging device 4 is as shown in FIG. 5 when the horizontal axis represents the plate width direction (X direction) and the vertical axis represents luminance.
When the luminance threshold value TH is equal to or higher than the set luminance threshold value TH or lower than the set luminance threshold value TL, a predetermined region corresponding to the threshold value is extracted as a cocoon candidate, and the X coordinate and Y coordinate thereof are stored in the candy candidate coordinate memory.
Note that the average luminance of the black and white image may be calculated, and the threshold values TH and TL may be set based on the average luminance.

FIG. 6A shows a map in which the cocoon candidates are plotted using the coordinates stored in the heel candidate coordinate memory. This is obtained by plotting the surface of a steel plate having a width of 800 mm over about 50 m in the longitudinal direction and extracting the wrinkle candidates extracted from the captured image.
The wrinkle candidates 1 and 2 confirmed at two places on the upper left of the map are aperiodic noise. Since there is no periodicity in the longitudinal direction (Y direction) and the positions in both the plate width directions (X direction) are different, it can be determined that there is no periodic defect.
On the other hand, the wrinkle candidates 3, 4 and 5 located at the top of the map are periodic wrinkles, and it can be confirmed that a wrinkle candidate is missing from the wrinkle candidates 4 and 5.
In addition, on the right side of the periodic tile, there are two or three rows of candidates in the longitudinal direction, and at this processing stage, it is not possible to determine whether it is a periodic defect or noise. It has also been confirmed that noise such as oil and water drops does not affect the quality of the product.

Step S12
A signal waveform is created from the coordinates of the extracted wrinkle candidate using a longitudinal coordinate function. That is, in the method shown in FIG. 1 (c), using the coordinates of the extracted wrinkle candidate, a portion with a wrinkle candidate is displayed as 1 and a portion without a wrinkle candidate is displayed as a rectangle, and the width of the rectangle is assumed. Created with a width of the period variation ΔY. In addition, although the rectangular wave was used, you may represent with a triangular wave or a trapezoid wave.

In this case, when processing a wrinkle candidate including obvious noise as shown in FIG. 6A, the signal waveform is created using all the coordinates of the wrinkle candidate and the autocorrelation function is calculated. It is difficult to detect the property, and the processing efficiency is also poor.
Therefore, the processing region 13 that is long in the longitudinal direction shown in FIG. 6A is selected, and a signal waveform is created using the coordinates of the wrinkle candidate included in the processing region 13.
In selecting the processing region 13, the image processing device 9 automatically selects the periodic tiles by utilizing the property that the periodic tiles have the same or neighboring X coordinates and the property that the Y coordinates have substantially equal intervals. Alternatively, the operator of the inspection apparatus may make a selection while referring to the map diagram of FIG.
In addition, without narrowing down as described above, the coordinate axis including the wrinkle candidate may be divided into a plurality of pieces in the plate width direction of the band-like body in advance, and the processing may be performed for each of the divided processing regions 13.
In this case, if there is a wrinkle candidate near the boundary of the divided processing region 13, periodicity may not be detected. Therefore, the selection of the plurality of processing regions 13 is shown in FIG. As described above, it is effective to divide the belt-shaped body with a lapping margin in the plate width direction.

Step S13
For the signal waveform created for each divided processing region 13, the autocorrelation function of the signal waveform is calculated, the peak obtained from the calculation result is detected, and the presence or absence of periodicity of the signal waveform is determined based on the peak To do.
The peak detection selects a peak whose height exceeds a preset threshold and has the minimum period among the assumed periods.
If periodicity is detected, an alarm is issued to the operator of the inspection apparatus.
If not detected, the process is performed for another process area 13 in step S12, and the processes in steps S12 and S13 are repeated until the inspection of all the process areas is completed.

Next, another embodiment of the image processing in the present invention will be described with reference to the flowchart shown in FIG.
Step S21
Since step S21 is the same as step S11, the description thereof is omitted.

Step S22
In this process, harmless flaws (groups) produced by transferring rolls (groups) or oil or water droplets adhering to the rolling rolls to the material to be rolled are in the shape, brightness, area, etc. within each group. Since it has high similarity, a process for removing noise such as harmless soot from the extracted soot candidates is performed using this property.

  As an identification method for removing noise, there are an identification method using a feature amount, a pattern matching identification method using an image, or an identification method using both of them.

The identification method using the feature amount is to measure the feature amount such as the position, shape, brightness, etc. of the extracted heel candidate, and compare the measured feature amount with a reference value, so that the heel candidate belongs to the roll heel group It is a method of identifying whether it belongs to the harmless group.
The feature quantity to be measured is one or a combination of two or more selected from the position of the wrinkle candidate, the width and length of the circumscribed rectangle, the ratio of the length to the width, the area of the wrinkle candidate, the peripheral length, the luminance, and the like.
As the reference value, data obtained by actual operation stored in the auxiliary storage device 8 may be used, or an apparent roll か ら from the candidate image displayed on the display device 7 by the operator of the inspection device. Alternatively, obvious noise may be selected as a sample, and a feature amount measured from the selected sample may be used as a reference value.

The pattern matching identification method using images is an identification in which a reference image (template) selected in advance and a test image are compared in density on a pixel-by-pixel basis to determine whether the image is similar to the reference image based on the degree of correlation. There are L1 norm and normalized correlation coefficient as an evaluation scale representing the degree of correlation.
When the L1 norm is used as the evaluation scale, the calculation time can be shortened by a calculation truncation method called a residual sequential test method. However, when the brightness of the reference image and the inspection image is different, a similar image is observed with the naked eye. Even so, the correlation value may be determined to be low.
On the other hand, when a normalized correlation coefficient is used as an evaluation scale, there is a feature that it is not affected by the difference in brightness between the reference image and the inspection image, but it has a feature that processing time is required.
As the reference image (template), a reference image obtained by actual operation stored in the auxiliary storage device 8 may be used, but the operator of the inspection device can select from the images of the eyelid candidates displayed on the display device 7. It is desirable to select obvious rolls or obvious noise as the reference image (template).
FIG. 8 is a diagram illustrating an example of a pattern matching identification method using an image. In this example, the operator of the inspection apparatus selects (a) as a clear roll wrinkle from the wrinkle candidate images displayed on the display device 7, and selects (b) to (h). Among the seven inspection images, four images (b), (d), (f), and (g) are determined to match the reference image (a) (high correlation).

Step S23-24
Steps S23 to S24 are the same as steps S12 to S13, and a description thereof will be omitted.

It is a figure which shows the principle of the roll wrinkle detection method which concerns on this invention. It is the schematic of a surface flaw inspection apparatus. It is a block diagram of the principal part of a surface flaw inspection apparatus. It is a flowchart which shows one form of the image processing in this invention. It is a figure which shows the extraction process of a cocoon candidate. It is the map figure which plotted the extracted eyelid candidate. It is a flowchart which shows the other form of the image processing in this invention. It is a figure which shows one Example of the pattern matching identification method using an image.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Strip steel plate 2 Conveyance roller 3 Pulse generator 4 Imaging device 5 Input device 6 Graphic board 7 Display device 8 Auxiliary storage device 9 Image processing device 10 CPU
11 Main storage device 12 Data input / output control device 13 Processing area

Claims (10)

  In a roll-shaped roll wrinkle detection method for detecting a roll wrinkle by imaging the surface of a moving band and processing the captured image to detect roll wrinkles, a plurality of wrinkle candidates appearing at intervals in the longitudinal direction of the belt-shaped body are captured from the captured image. From the extracted coordinates of the wrinkle candidates, a signal waveform having information on the number of wrinkle candidates and the interval in the longitudinal direction and having a periodic fluctuation width of the roll wrinkle is created, and the autocorrelation function of the signal waveform Detecting a roll wrinkle by detecting a roll wrinkle by detecting a peak obtained from the result of the operation and determining the presence or absence of periodicity of the signal waveform based on the peak Method.   In a roll-shaped roll wrinkle detection method for detecting a roll wrinkle by imaging the surface of a moving band and processing the captured image to detect a roll wrinkle, a plurality of wrinkle candidates appearing at intervals in the longitudinal direction of the belt-shaped body Extraction, removing the noise from the extracted candy candidates, selecting the candy candidates, and having the information on the number of candy candidates and the longitudinal interval from the coordinates of the selected candy candidates, and assuming the periodic fluctuation of the roll candy By creating a signal waveform having a width, calculating an autocorrelation function of the signal waveform, detecting a peak obtained from the result of the calculation, and determining the presence or absence of periodicity of the signal waveform based on the peak, A method for detecting roll wrinkles of a band-shaped body, characterized by detecting roll wrinkles.   The method for detecting roll wrinkles of a band according to claim 2, wherein an identification method using a feature value is used for selecting the wrinkle candidates.   The method for detecting wrinkles on a strip according to claim 2, wherein a pattern matching method using an image is used for selecting the wrinkle candidates.   The method for detecting roll wrinkles of a band according to claim 2, wherein both the identification method using a feature amount and the pattern matching method using an image are used for selecting the wrinkle candidates.   The strip according to any one of claims 1 to 5, wherein the peak detection selects a peak whose height exceeds a preset threshold and has a minimum period among the assumed periods. Roll wrinkle detection method.   7. A series of processing from creation of the signal waveform to determination of the presence / absence of periodicity of the signal waveform is processed for each processing region divided into a plurality in the plate width direction of the belt-like body. The method for detecting roll wrinkles of a band according to any one of the above.   The method for detecting roll wrinkles of a band according to claim 7, wherein the plurality of processing regions are divided with a lapping margin in the plate width direction of the band.   Create a signal waveform to be used as an input signal for the autocorrelation function in the roll corrugation detection method using the autocorrelation function that uses the autocorrelation function to detect the roll wrinkle by imaging the surface of the moving band and processing the captured image. In the method, using the coordinates of the wrinkle candidate extracted from the captured image, a part having a wrinkle candidate is created with a rectangular wave having a height of 1 and a part having no wrinkle candidate is formed with a height of 0, and a roll that assumes the width of the rectangle A method for creating a signal waveform, characterized in that the period fluctuation width of the ridge is set.   In a roll-shaped roll wrinkle detection device that picks up an image of the surface of a moving belt-shaped body and detects a roll wrinkle by performing image processing on the captured image, an imaging device that images the surface of the band-shaped body is spaced apart in the longitudinal direction of the band-shaped body Means for extracting a plurality of wrinkle candidates appearing from the captured image, and information on the number of wrinkle candidates and the interval in the longitudinal direction from the coordinates of the extracted wrinkle candidates, and having an assumed periodic variation width of the roll wrinkles Means for creating a signal waveform, means for calculating an autocorrelation function of the signal waveform, means for detecting a peak obtained from the result of the calculation, and determining the presence or absence of periodicity of the signal waveform based on the peak An apparatus for detecting roll wrinkles of a belt-like body, comprising: an image processing apparatus having means.

Images