JP4516884B2 - Periodic defect inspection method and apparatus - Google Patents

Description

  The present invention relates to a periodic defect inspection method and apparatus for inspecting periodic defects on the surface of an object to be inspected.

  Conventionally, various techniques for inspecting periodic defects on the surface of an object to be inspected are known. For example, Patent Document 1 discloses a technique for detecting a continuous defect that occurs over the entire length of a steel sheet in a rolling line for producing a strip of the steel sheet. According to this, the number of defects existing at the same position in the width direction of the steel sheet is accumulated in the longitudinal direction, the position in the width direction where the number exceeds a predetermined threshold is detected, and the same is true for a plurality of steel sheets. Continuous defects are determined by examining defects at positions in the width direction.

  In Patent Document 2, the feature amount of a defect existing at the same position in the longitudinal direction of the steel sheet is evaluated, and the position in the width direction is also recorded in the case of a type of defect having periodicity. Next, when a periodic defect of the same type is detected at another position in the longitudinal direction, the width direction position is compared with the recorded width direction position. Recognizing that the defect is a periodic defect, the defect interval is obtained from the longitudinal position of both. The periodic defect interval obtained first is recorded as a reference for this type of periodic defect, and then when the defect interval is obtained from the same type of periodic defect, the periodicity is compared with this reference. It is determined whether it is a defect interval. Each time a periodic defect interval is detected, an evaluation value determined for each type is added. When the cumulative evaluation value exceeds a threshold value, an alarm is output and all records are reset.

  Further, in Patent Document 3, the defect intervals of the defects detected at the same position in the width direction are totaled and determined as a periodic defect when the standard deviation of the defect interval is small.

  Patent Document 4 discloses a technique for extracting periodic defects by directly analyzing image data on the surface of an object to be inspected.

JP 2004-125686 A JP-A-7-198627 JP-A-6-294759 JP 2001-281154 A

  However, in the periodic defect inspection according to the conventional technique, it is not easy to detect the periodic defect by low load calculation when the object to be inspected is displaced in the width direction perpendicular to the traveling direction. That is, in the periodic defect inspection according to Patent Documents 1 and 3, the possibility that the steel sheet is displaced in the width direction perpendicular to the traveling direction when traveling between the rolls is not considered. For example, according to the embodiment of Patent Document 1, after creating a histogram of defect distribution states in the width direction at each position in the longitudinal direction, each histogram is divided into pixel units of about 50 to 60 mm in the width direction, Whether or not the defect is a continuous defect is determined by comparing the value obtained by integrating the defect data in the longitudinal direction for each division unit with a threshold value (see paragraphs 0023 and 0024). Therefore, when the steel plate is displaced in the width direction and continuous defects are distributed in a plurality of divided units and spread over a wide range, detection becomes difficult. Also in Patent Document 3, when detecting a periodic defect, the position in the width direction of the defect of the steel plate is determined by fixing it in units of pixels, and when the steel plate is displaced in the width direction, an image is taken with a camera. The pixel position in the width direction of the periodic defect is shifted and the periodicity is not detected.

  The periodic defect inspection according to Patent Document 2 corresponds to the displacement in the width direction of the object to be inspected, but the periodic defect interval is individually evaluated when determining the periodic defect. That is, the defect interval first detected for the same type of periodic defect is recorded as a reference, and each time the periodic defect interval is detected thereafter, a determination is made in comparison with this reference. The determination of the periodicity becomes inaccurate. For example, (1) when there are many noises that are mistakenly recognized as defects, and there is a noise-defect interval or a noise-noise interval that is substantially equal, an over-detection is made as a periodic defect. (2) It is described that even when a part of the periodic defect of the object to be inspected is missing and missing, it can be detected by dividing and comparing the recorded defect intervals (paragraph 0020). However, in the case of division comparison, if there is an error, calculation may not be performed properly. (3) Also, when detecting one periodic defect, if one missing piece or two missing pieces continue, the period is detected as a double or triple value. (4) Since a memory is reset when a periodic defect alarm is output, even if a periodic defect of the same type is subsequently detected, it is not immediately determined as a periodic defect.

  Furthermore, there is a conventional technique as described in Patent Document 4, but in such a periodic defect inspection, since the image data signal obtained is directly processed, the calculation load is large. It is difficult to add as a part of processing of a normal defect inspection apparatus.

  The present invention has been made in view of the above problems, and when the object to be inspected is displaced in the width direction perpendicular to the traveling direction, the periodic defect can be detected by low load calculation. It is an object of the present invention to provide a periodic defect inspection method and apparatus capable of detecting the periodicity more accurately.

  In order to solve the above problems, according to the present invention, a defect classification step of classifying defects detected in image data obtained by imaging an object to be inspected into periodic defect candidates and aperiodic defects; , And included in each of the plurality of rectangular periodicity determination areas that are arranged in the width direction of the object to be inspected so as to cover the entire image data, and whose long side direction coincides with the longitudinal direction of the object to be inspected. A defect candidate counting step for counting each of the periodic defect candidates, and when the number of the periodic defect candidates exceeds a first threshold, adjacent periodic defect candidates for the periodicity determination region A histogram creation step of creating a histogram having the horizontal defect interval in the longitudinal direction and the frequency in the vertical axis, and each frequency within the predetermined range of the horizontal axis of the histogram corresponding to the frequency Axis value There is a frequency that exceeds the second threshold in the processed histogram obtained by adding weighted values of K times (K = 2, 3,..., N (N is a predetermined natural number)). In this case, there is provided a periodic defect inspection method characterized by including a periodic defect determination step for determining that a periodic defect having a period on the horizontal axis is included.

  Further, in the periodic defect inspection method, the periodicity determination region covers the first layer arranged in close contact with each other so as to cover the entire image data, and the entire image data. The long sides of the periodicity determination region of the first layer are arranged in the second layer, the long sides of which are arranged close to each other, and the long side of the periodicity determination region of the second layer is the length of the periodicity determination region of the second layer. You may arrange | position so that it may correspond to the centerline of a direction.

  In order to solve the above-described problem, according to the present invention, an imaging unit that images the surface of an object to be inspected traveling between a plurality of rolls, and image data of the surface of the object to be inspected obtained by the imaging unit. And a storage means for storing a defect included in the image data, classifying the detected defect into a periodic defect candidate and an aperiodic defect, and covering the entire image data The periodic defect candidates included in each of the plurality of rectangular periodicity determination areas that are arranged in the width direction of the object to be inspected and whose long side direction coincides with the longitudinal direction of the object to be inspected are counted. When the number of periodic defect candidates exceeds the first threshold, the longitudinal defect interval between adjacent periodic defect candidates is plotted on the horizontal axis and the frequency is plotted on the vertical axis. Create a histogram with axes , Each frequency within the predetermined range of the horizontal axis of the histogram is multiplied by K times the horizontal axis corresponding to the frequency (K = 2, 3,..., N (N is a predetermined natural number)) A period for determining that a periodic defect having a value on the horizontal axis as a period is included when there is a frequency that exceeds the second threshold in the obtained processing histogram by adding the weights of the obtained values with weights. There is provided a periodic defect inspection apparatus characterized by having a sex determination means.

  Further, in the periodic defect inspection apparatus, the periodicity determination region covers the first layer arranged in close contact with each other so as to cover the entire image data, and the entire image data. The long sides of the periodicity determination region of the first layer are arranged in the second layer, the long sides of which are arranged close to each other, and the long side of the periodicity determination region of the second layer is the length of the periodicity determination region of the second layer. You may arrange | position so that it may correspond to the centerline of a direction.

  According to the present invention, it is possible to detect a periodic defect by low load calculation when the object to be inspected is displaced in the width direction perpendicular to the traveling direction. In addition, the periodic defect and its periodicity can be detected more accurately.

  Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the drawings. In the present specification and drawings, elements having substantially the same functional configuration are denoted by the same reference numerals, and redundant description is omitted.

  FIG. 1 is a configuration diagram of a periodic defect inspection apparatus 1 in which a periodic defect inspection according to an embodiment of the present invention is applied to a typical steel rolling line. The steel rolling line includes an unwinding reel 10 as a starting point of the line, a take-up reel 14 as an end point, a transport roll 12 that advances a steel plate 11 that is an inspection object between those reels, and a rolling roll 13 that performs a rolling process. Consists of. An illumination 15 and a CCD line camera 16 as an imaging means are disposed between the rolling roll 13 and the take-up reel 14 so that light emitted from the illumination 15 is reflected by a steel plate and enters the CCD line camera 16. Has been. An image memory 17 connected to an image data processing device 18 is connected to the CCD line camera 16, and the image data processing device 18 is further connected to a display 19.

  Next, a periodic defect inspection method according to an embodiment of the present invention will be specifically described using the periodic defect inspection apparatus. The strip-shaped steel plate 11 delivered from the unwinding reel 10 is advanced in the direction of the arrow by the transport roll 12, rolled by the rolling roll 13, and then taken up by the take-up reel 14. The periodic defect inspection on the surface of the steel sheet is performed by illuminating the rolled steel sheet 11 with light from the illumination 15 and imaging the surface with the CCD line camera 16.

  As shown in FIG. 2, the CCD line camera 16 at a fixed position continuously captures the steel plate 11 traveling in the longitudinal direction Y at a predetermined time interval, captures the line image 20 in the width direction X of the steel plate 11, It memorize | stores in the image memory 17 which is a memory | storage means. The periodicity determination device 18 combines these line images 20 captured continuously in the longitudinal direction Y of the steel plate 11 by a predetermined number of connections, and the length of the direction X as shown in FIG. Image data 21 having the width a and the length Y in the longitudinal direction Y is obtained. Here, in the same image data 21, the image data 21 so that a predetermined minimum number n (n is a natural number of 2 or more) of periodic defects periodically continuing in the longitudinal direction Y of the steel plate 11 can be included. It is necessary to set the length a in the longitudinal direction Y to be larger than n times the maximum value (maximum outer peripheral length) of the outer peripheral lengths of various rolling rolls 13 on the steel rolling line. This is because periodic defects are generated by the rolling roll 13 and the period becomes the outer peripheral length of the rolling roll. For example, as shown in FIG. 2, when the rolling roll 13 having an outer peripheral length L has a flaw occurrence cause 22, after the flaw occurrence cause 22 comes into contact with the steel plate 11 to generate a periodic defect 23, the rolling roll 13 Rotates once again, and comes into contact with the steel plate 11 again to generate the next periodic defect 24. The defect interval between these periodic defects 23, 24, that is, the period is equal to the outer peripheral length L of the rolling roll 13. Therefore, in order to include at least n periodic defects having the period L in the image data 21, it is necessary to set the length a in the longitudinal direction Y of the image data 21 to be larger than n times the outer peripheral length L. . For example, as shown in FIG. 1, when a rolling roll 13 having an outer peripheral length L and a rolling roll 13 having an outer peripheral length M are arranged on a steel rolling line, and L <M, n is set to about 10 and an image If the length of the data 21 in the longitudinal direction a is about 10 times the maximum outer peripheral length M, both the generation of wrinkles by the outer peripheral length L of the rolling roll 13 and the generation of wrinkles by the outer peripheral length M of the rolling roll 13 Can be detected without omission.

  Next, the procedure of the periodic defect inspection process after the image data 21 is taken into the periodicity determination device 18 will be described with reference to FIG. After the image data 21 of the surface of the object to be inspected is captured (SP0), the captured image data 21 is detected for defects and its features by binarization or the like (SP1). For example, by performing defect detection processing such as binarization on the image data 21 obtained by imaging the steel plate 11, the feature-detected image data 25 shown in FIG. 5 is obtained. The feature-detected image data 25 includes, for example, a periodic defect α, a large-area non-periodic defect β, a small-area non-periodic defect γ, and a small-area collective non-periodic defect δ. The periodic defect α is a defect which is caused by the rolling roll 13 and has a period of the outer peripheral length L of the rolling roll 13, and the area thereof is the same as or smaller than the defect γ or δ. When the defects α, β, γ, and δ are detected, information on characteristics such as the area, length, shape, and position on the image data 21 of the defects α, β, γ, and δ is also acquired. .

  The detected defects are classified into periodic defect candidates and aperiodic defects according to their characteristics (SP2). For example, a defect whose area size exceeds a predetermined value can be classified as an aperiodic defect. In the case of the feature-detected image data 25 shown in FIG. 5, the defect β having a large area is classified as an aperiodic defect, and α, γ, and δ having a small area are classified as periodic defect candidates. By this classification, the non-periodic defect β is removed, and the classified image data 26 shown in FIG. 6 is obtained. In the classification, in addition to the defect area, various characteristics such as the length, shape, and position on the image data 25 may be used.

  In addition, a plurality of rectangular periodicity determination areas 27 are defined as follows for the image data 26 having a predetermined size obtained as described above. As shown in FIG. 7, the plurality of periodicity determination areas 27 are divided and arranged in the width direction X of the steel plate 11 so as to cover the entire classified image data 26. The long side of each periodicity determination area 27 has the direction coincident with the longitudinal direction Y of the steel plate 11, and the length is set to be the same as or slightly larger than the length a of the classified image data 26. ing. In addition, as shown in FIG. 7, the width direction of each periodicity determination region 27 can be included so that a periodic defect can be included even if the steel plate 11 is displaced in the width direction X when passing the rolling roll 13. The length l is set larger than the displacement amount x in the width direction X of the steel plate 11.

  Further, in this embodiment, the first layer 28 in which the periodicity determination region 27 is arranged in close contact with each other so as to cover the entire classified image data 26, and the entire classified image data 26. And the second layer 29 in which the periodicity determination region 27 is arranged so that the long sides are in close contact with each other so as to overlap each other. For example, a first layer 28 that covers the entire classified image data 26 and a second layer 29 that covers the entire classified image data 26 are shown in FIG. The first layer 28 is provided with a periodicity determination region 27 so that the long sides are in close contact with each other. Similarly, the long side of the second layer 29 is also in close contact with each other. A periodicity determination area 27 is arranged. Here, the long sides AB and CD of the periodicity determination area ABCD constituting the first layer 28 are respectively aligned with the longitudinal center lines of the periodicity determination areas EFGH and HGJI constituting the second layer 29. Match. That is, the periodicity determination region 26 of the first layer 28 is arranged so that its long side coincides with the longitudinal center line of the periodicity determination region 27 of the second layer 29.

  As described above, since the periodicity determination region 27 has a two-layer structure in which the center line is shifted, the periodic defect generated from the same flaw occurrence factor in the periodicity determination region 27 of any layer. Can be included. For example, in the case of FIG. 7, the periodicity determination area ABCD of the first layer 28 can include only 32 of the periodic defects 30, 31, and 32, but the second layer HGJI includes 30, 31 and All 32 can be included.

Subsequently, periodic defect candidates included in each of a plurality of periodicity determination areas 27 determined so as to cover the entire classified image data 26 from defects classified as periodic defect candidates in SP2. Is counted (SP3). Here, among the plurality of periodicity determination areas 27 defined for the image data 26 classified as shown in FIG. 7, any three periodicity determination areas 27a, 27b and 27c shown in FIG. Explained. The periodicity determination region 27a includes the periodic defects α _{1 to} α ₁₄ and the non-periodic defect γ generated from the cause 22 of wrinkles of the rolling roll 13. The periodicity determination region 27b includes non-periodic defects δ _{1 to} δ ₉ . The periodicity determination region 27c includes non-periodic defects ε _{1 to} ε ₃ . That is, when the periodic defect candidates included in the periodicity determination areas 27a, 27b, and 27c are counted, the periodicity determination area 27a is 15, the periodicity determination area 27b is 9, and the periodicity determination area 27c is 3. It has become.

Next, for each of the periodicity determination areas, the number of periodic defect candidates counted in SP3 is compared with a first threshold value, and if this number exceeds the first threshold value, the periodicity determination is performed. For a region, a histogram is created with the horizontal axis representing the defect interval in the longitudinal direction between adjacent periodic defect candidates and the vertical axis representing the frequency (SP4). For example, the periodicity determination areas 27a, 27b, and 27c shown in FIG. 8 include 15, 9, and 3 periodic defect candidates, respectively. In FIG. 8, when the first threshold value for determination is 8, the number of defects included in each of the periodicity determination regions 27a and 27b exceeds this threshold value. Therefore, the periodic defect candidates α _{1 to} α ₁₄ , γ and δ _{1 to} δ ₉ included in these periodicity determination areas 27a and 27b are determined to be periodic defects, and will be described later. A histogram is created. On the other hand, the number of periodic defect candidates included in the periodicity determination region 27c is less than this threshold value. Therefore, it is determined that the periodicity determination region 27c includes only a non-periodic defect, and a histogram is not created.

FIG. 9 shows a histogram actually created based on the periodic defect candidates included in the periodicity determination region 27a shown in FIG. FIG. 11 shows a histogram actually created based on the periodic defect candidates included in the periodicity determination region 27b shown in FIG. The histogram shown in FIG. 9 is obtained from the positional relationship between the defects α _{1 to} α ₁₄ and γ in the periodicity determination region 27a shown in FIG. 8 (see FIG. 10). A histogram is created with the horizontal axis representing the defect interval in the longitudinal direction Y between adjacent defects and the vertical axis representing the frequency thereof. In this histogram, the frequency of data having the period L of the periodic defects α _{1 to} α ₁₄ as the defect interval is 12, and a remarkable peak is formed. On the other hand, the histogram shown in FIG. 11 is obtained from the positional relationship (see FIG. 12) of the defects δ _{1 to} δ ₉ in the periodicity determination region 27b shown in FIG. A histogram is created with the horizontal axis representing the defect interval in the longitudinal direction Y between adjacent defects and the vertical axis representing the frequency thereof. Since these defects are non-periodic defects, the frequency of the defect intervals of each data is almost evenly distributed in this histogram.

In the case described with reference to FIGS. 9 and 10, the defect occurred without omission, but the case of missing teeth is also conceivable. For example, FIG. 13 is a diagram assuming a case where α ₂ , α ₃ , α _5, and α ₉ are missing from the periodic defect interval shown in FIG. Since α ₂ and α ₃ are missing, the defect interval between α ₁ and α ₄ is three times the period L, and since α ₅ and α ₉ are missing, the defect interval between α ₄ and α ₆ and α defect spacing ₈ and alpha ₁₀ is twice the period L. Similar to the histogram of FIG. 9 obtained from the positional relationship shown in FIG. 10, in the histogram shown in FIG. 13, data having a defect interval of the period L of the periodic defects α _{1 to} α ₁₄ forms a peak. However, the distribution also spreads in the periods 2L and 3L due to missing teeth, and the peak is lower than in the case of the histogram shown in FIG.

  Therefore, with respect to the histogram obtained at SP4, the horizontal axis value of each frequency is multiplied by K times (K = 2, 3,..., N ( N is a predetermined natural number)). By adding the frequency of the value data with weighting, correction processing is performed to cope with a case where a part of the periodic defect is missing, and a processing histogram is obtained. . Then, in this processed histogram, when there is a frequency exceeding the second threshold, it is determined that a periodic defect whose period is the value on the horizontal axis is included (SP5). Here, the minimum value and the maximum value of the predetermined range determine the minimum value and the maximum value that can be actually taken by the period value.

  This will be specifically described with reference to FIG. Assuming that the minimum value and the maximum value of the outer peripheral length of the plurality of rolling rolls 13 that cause periodic defects are (1/2) L and (3/2) L, respectively, the predetermined minimum value is (1/2) L The predetermined maximum value is defined as (3/2) L. Here, the outer peripheral length of the rolling roll 13 is set to the minimum value and the maximum value in the predetermined range, but the predetermined minimum value and the predetermined maximum value may be determined based on other values.

In the histogram shown in FIG. 13, the range in which the horizontal axis is (1/2) L to (3/2) L times K (K = 2, 3,... N (N is a predetermined natural number)) is , L to 3L, (3/2) L to (9/2) L, ..., (N / 2) L to (3N / 2) L, respectively. In the case of FIG. 13, since N = 4, three ranges of L to 3L, (3/2) L to (9/2) L, and 2L to 6L are considered. Next, after assigning a weight of 0.8 ^(K-1) to the frequency of data in these ranges, the horizontal axis represents the frequency of data in the range of (1/2) L to (3/2) L. to add. In other words, the frequency of data in the range of L to 3L on the horizontal axis is multiplied by 0.8 and added to the frequency of data in the range of (1/2) L to (3/2) L. The horizontal axis is (3/2) L~ (9/2) and 0.8 ^twice the frequency of the data in the range L, (1/2) L~ (3/2 ) in the range of L Is added to the data frequency. Similarly, the horizontal axis is 0.8 ³ times the frequency of the data in the range of 2L～6L, it is added to the frequency of the data within the range of (1/2) L~ (3/2) L . The processing histogram thus obtained is shown in FIG.

As apparent from the processing histogram shown in FIG. 15, even when a part of the periodic defect is missing, it is possible to cope with the missing tooth correction described above. Since the frequency of the periodic defect that is a natural number multiple of the true period caused by the missing tooth is weighted and added to the frequency of the true period, the periodicity determination accuracy of the periodic defect is improved. The weighting at this time is performed so that the true period finally becomes the largest value. Here, weighting is performed by multiplying the data to be added by 0.8 ^(K-1) , but this weighting may be performed by multiplying the data to be added by a ^(K-1) (a is 0). <A <1 real number) and other methods may be used. The processed histogram shown in FIG. 16 is obtained by performing the same missing tooth correction as described above on the histogram shown in FIG.

  Next, in the obtained processed histogram data, when there is a frequency exceeding the second threshold, the periodicity determination region has the frequency on the horizontal axis, that is, the defect interval as a cycle. It is determined that a periodic defect is included. More specifically, in the processed histogram shown in FIG. 15, there is a frequency 7.24 exceeding the second threshold 6, so the periodicity determination region 27 a has a value on the horizontal axis of this frequency, that is, a defect interval L It is determined that it contains a periodic defect with a period. On the other hand, in the processed histogram shown in FIG. 16, the highest frequency in the processed histogram is 1.44, and there is no frequency exceeding the second threshold value 4.8. It is determined that no periodic defect is included. The second threshold was calculated as 0.6 × (sum of all frequencies of the SP4 histogram).

The information obtained in SP1 to SP5 is collected, and information indicating whether the defect is a periodic defect or a non-periodic defect is displayed on the display 19 (SP6). Furthermore, information on each defect obtained in SP2 is also displayed. In addition, when the detected defect is a periodic defect, the period obtained in SP5 is also displayed. More specifically, when the defects α _{1 to} α ₁₄ are detected and displayed, information indicating that the defect is a periodic defect having a period L is displayed together with other characteristic information. On the other hand, when the defects β, γ, and δ _{1 to} δ ₉ are detected and displayed, information indicating that the defect is not a periodic defect and other characteristic information are displayed.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, this invention is not limited to the example which concerns. It is obvious for those skilled in the art that various changes or modifications can be conceived within the scope of the technical idea described in the claims. It is understood that it belongs to.

  For example, in the above-described embodiment, the case where a steel plate is used as the object to be inspected has been described. However, the object to be inspected may be paper, a film, a non-woven fabric, or the like.

  In the above-described embodiment, the case where the object to be inspected is a sheet shape has been described.

  In the above-described embodiment, the case where the inspection determination result is output to the display 19 has been described. However, the output destination may be an output device to some medium such as a printing device or a disk recording device. However, it may be a device for performing another process, such as an analysis device or a transmission device.

  In the above-described embodiment, the processing procedure of SP1 to SP6 has been described. However, even if the order is changed within a range where the substantial processing content does not change, it does not depart from the present invention.

  In the above-described embodiment, the periodic defect inspection is performed through SP1 to SP6. However, the periodic defect inspection may be performed only in a part of the process.

A histogram was created from the image data obtained by imaging the surface of the steel sheet under the following conditions.
1) Minimum value of the predetermined range in SP5: 500 mm
2) Maximum value of the predetermined range in SP5: 2500 mm
3) Length of side in width direction of periodicity determination region: 20 mm
4) Histogram creation pitch: 50 mm
5) Histogram horizontal axis maximum value: 20000 mm

  The obtained histogram is shown in FIG. In this case, there is almost no loss of periodic defects, and the number of defects in the periodicity determination area is 46. Therefore, the sum of the frequency of defect intervals is 45.

  FIG. 18 shows the histogram of FIG. 17 normalized by the sum 45 of the frequency of defect intervals. Since there is almost no missing tooth of the periodic defect, a remarkable peak is formed. At this time, it is found that the period of the periodic defect is about 1800 mm.

In the histogram of FIG. 18, each frequency within the range of 500 mm to 2500 mm is a value obtained by multiplying the value of the horizontal axis by K times (K = 2, 3,..., N (N is a predetermined natural number)). FIG. 19 shows the result of adding the frequency of .times.0.8 ^(K-1) times. This tooth loss correction forms peaks at 1/2 and 1/3 of the period, but does not exceed the actual period.

A histogram was created from the image data obtained by imaging the surface of the steel sheet under the following conditions.
1) Minimum value of the predetermined range in SP5: 500 mm
2) Maximum value of the predetermined range in SP5: 2500 mm
3) Length of side in width direction of periodicity determination region: 20 mm
4) Histogram creation pitch: 50 mm
5) Histogram horizontal axis maximum value: 20000 mm

  The obtained histogram is shown in FIG. In this case, there are missing teeth of periodic defects, and the number of defects in the periodicity determination region is 36. Accordingly, the sum of the frequency of defect intervals is 35.

  FIG. 21 shows the histogram of FIG. 20 normalized by the sum 35 of the frequency of defect intervals. Due to the missing teeth of periodic defects, multiple peaks were formed, and the true period could not be determined.

In the histogram of FIG. 21, each frequency in the range of 500 mm to 2500 mm is a value obtained by multiplying the value of the horizontal axis by K times (K = 2, 3,..., N (N is a predetermined natural number)). FIG. 22 shows the result of adding the frequency of data multiplied by 0.8 ^(K−1) . By this tooth loss correction, a peak of the maximum frequency appeared at a true period of about 1800 mm. The value of the maximum frequency exceeded the second threshold value 0.5, and was determined to be a periodic defect.

FIG. 23 shows weighting of 1.0 ^(K-1) times instead of 0.8 ^(K-1) times when correcting missing teeth on the histogram of FIG. In this case, a peak exceeding the true period appears. Therefore, it can be seen that a value of about 0.8 is appropriate for weighting.

  According to the present invention, it is possible to inspect periodic defects on the surface of an object to be inspected such as a steel plate.

It is explanatory drawing of the typical steel rolling line to which the periodic defect inspection which concerns on embodiment of this invention is applied. It is explanatory drawing of continuous imaging of the line image of the steel plate by a CCD line camera. It is a figure explaining the image data of this invention. 2 is a flowchart of a preferred embodiment of the present invention. It is a figure explaining the image data by which the feature detection was carried out. It is a figure explaining the image data from which the defect was detected. It is a figure explaining the periodicity determination area | region of this invention. It is a figure explaining the relationship between the periodicity determination area | region of this invention, and the candidate of a periodic defect. It is a histogram based on the defect in a periodicity determination area | region. It is a figure explaining the positional relationship of the defects in a periodicity determination area | region. It is a histogram based on the defect of a periodicity determination area | region. It is a figure explaining the positional relationship of the defects in a periodicity determination area | region. It is a histogram based on the defect of a periodicity determination area | region. It is a figure explaining the positional relationship of the defects in a periodicity determination area | region. FIG. 14 is a processing histogram after tooth missing correction is performed on the histogram of FIG. 13. FIG. 12 is a processed histogram after tooth missing correction is performed on the histogram of FIG. It is a histogram based on a periodic defect when there is no missing tooth. It is the histogram which normalized the histogram of FIG. FIG. 16 is a processing histogram obtained by correcting the missing tooth from the histogram of FIG. 15. FIG. It is a histogram based on a periodic defect when there is a missing tooth. 21 is a histogram obtained by normalizing the histogram of FIG. 20. FIG. 22 is a processed histogram obtained by correcting the missing tooth in the histogram of FIG. 21 (weighting formula: 0.8 ^(K−1) ). FIG. 22 is a processed histogram obtained by correcting the missing teeth in the histogram of FIG. 21 (weighting formula: 1.0 ^(K−1) ).

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Periodic defect inspection apparatus 10 Unwinding reel 11 Plate-shaped steel plate 12 Conveyance roll 13 Roll roll 14 Take-up reel 15 Illumination 16 CCD line camera 17 Image memory 18 Image data processing apparatus 19 Display 20 Line image 21, 25, 26 Image Data 22 Cause of wrinkle generation 23, 24 Periodic defect 27, 27a, 27b, 27c Periodicity determination area 28, 29 Layer of periodicity determination area L, M Roll perimeter length X Width direction of steel sheet Y Advancing direction of steel sheet A, B , C, D, E, F, G, H, I, J, K, L Vertex of periodicity determination area a Length in longitudinal direction of image data l Length in width direction of periodicity determination area α, α ₁ to α ₁₄ periodic defect β non-periodic defect with large area γ, δ _{1 to} δ ₉ , ε _{1 to} ε ₃ aperiodic defect

Claims (4)

A defect classification step for classifying defects detected in image data obtained by imaging the object to be inspected into periodic defect candidates and non-periodic defects;
The plurality of rectangular periodicity determination regions that are divided and arranged in the width direction of the object to be inspected so as to cover the entire image data and that have a long side direction that coincides with the longitudinal direction of the object to be inspected A defect candidate counting step for counting the number of candidates for periodic defects,
When the number of the periodic defect candidates exceeds a first threshold, the longitudinal defect interval between adjacent periodic defect candidates is plotted on the horizontal axis and the frequency is plotted on the vertical axis. A histogram creation process for creating a histogram
Each frequency within the predetermined range on the horizontal axis of the histogram is multiplied by K times (K = 2, 3,..., N (N is a predetermined natural number)) corresponding to the frequency. In the processing histogram obtained by adding the frequency of values with weights, when there is a frequency that exceeds the second threshold, the periodic defect that is determined to include a periodic defect with the value on the horizontal axis as the period A periodic defect inspection method comprising: a determination step. The periodicity determination region is arranged such that the long sides are closely arranged so as to cover the entire image data, and the first layer arranged so that long sides are closely attached so as to cover the entire image data. And arranged so that the long side of the periodicity determination region of the first layer coincides with the longitudinal center line of the periodicity determination region of the second layer. The periodic defect inspection method according to claim 1, wherein: An imaging means for imaging the surface of the object to be inspected traveling between a plurality of rolls;
Storage means for storing image data of the surface of the object to be inspected obtained by the imaging means;
Detecting defects included in the image data, classifying the detected defects into periodic defect candidates and aperiodic defects, and in the width direction of the inspection object so as to cover the entire image data The periodic defect candidates included in each of the plurality of rectangular periodicity determination areas that are arranged in a plurality of rectangular shapes whose long side directions coincide with the longitudinal direction of the inspected object are divided, and the periodic defect candidates are counted. When the number exceeds the first threshold, a histogram is created for the periodicity determination region, with the horizontal defect interval between adjacent periodic defect candidates and the vertical axis indicating the frequency. , Each frequency within the predetermined range of the horizontal axis of the histogram is multiplied by K times the horizontal axis corresponding to the frequency (K = 2, 3,..., N (N is a predetermined natural number)) Processing histogram obtained by weighting and adding frequency A periodic defect inspection device characterized by comprising periodicity determining means for determining that a periodic defect having a value on the horizontal axis is included when a frequency exceeding a second threshold is present in the ram. . The periodicity determination region is arranged such that the long sides are closely arranged so as to cover the entire image data, and the first layer arranged so that long sides are closely attached so as to cover the entire image data. And arranged so that the long side of the periodicity determination region of the first layer coincides with the longitudinal center line of the periodicity determination region of the second layer. The periodic defect inspection apparatus according to claim 3, wherein:

Images