JPH079408B2 - Surface defect detector - Google Patents

Description

TECHNICAL FIELD The present invention relates to an apparatus for detecting a surface defect of a test material,
For example, the present invention relates to a device for detecting flaws and the like on the surface of rolled steel.

[Prior art]

For example, on the surface of the steel material after rolling, there may be a defect called "hege", in which thin steel pieces are attached. Conventionally, such a flaw has been detected by visual inspection by an inspector who stands by on the production line.

[Problems to be Solved by the Invention]

However, such visual inspection requires skill because it targets the steel material conveyed on the production line, and also puts a heavy burden on the inspection staff.

Therefore, attempts have been made to automatically detect surface defects by imaging the surface of a steel material using a TV camera or the like and performing image processing. In other words, although the high temperature steel material emits light by itself, it is intended to detect defects by paying attention to the fact that, in the flaw portion, the recuperation after cooling becomes slower than that in other portions.

There were various problems in the practical application of this type of defect detection device, but the applicants of the present application filed and have already made public notices of Japanese Patent Publication Nos. 61-18694, 57-52983, 57-57- Since these problems have been successively solved by the technique disclosed in Japanese Patent Application No. 48735 and Japanese Patent Application No. 63-261548, which is pending, efforts have been focused on further improving the defect detection accuracy.

Therefore, it is an object of the present invention to provide a surface defect detection device that detects defects on the surface of a test material with high accuracy.

[Means for Solving the Problems]

In order to achieve the above object, the surface defect detection apparatus of the present invention images a surface of a test material (BM) moving in a predetermined direction,
The original image signal (O
(Z)) in which the main scanning direction x is a direction orthogonal to the moving direction and the sub-scanning direction z is parallel to the moving direction (61); main scanning direction for one main scanning line Storage unit (327) for holding the smoothed image signal (P (z)) addressed to each pixel; the original image signal (O (z)) of each pixel in the main scanning direction generated by the imaging unit (61); The smoothed image signal (P (z)) is smoothed in the sub-scanning direction using the smoothed image signal (P (z)) of the corresponding pixel of the storage means (327) up to the main scanning line immediately before the main scanning line to which it belongs. ) Is generated and is updated and stored in the storage means (327). Smoothing means (321 to 326, 32)
8); the difference (O (O (z)) between the original image signal (O (z)) and the smooth image signal (P (z)) of the storage means (327) up to the main scanning line immediately before the main scanning line to which it belongs. z) -P
(Z)) corresponding to pixels at the same position in the main scanning direction to generate a differential image signal (O (z) -P (z)).
1); and the differential image signal (O) generated by the processing means (321).
(Z) -P (z)) based on (z) -P (z)) to detect a defect on the surface of the test material (BM). In addition, in order to facilitate understanding, the reference numerals of corresponding elements in the embodiments shown in the drawings and described later are added in parentheses for reference.

[Action]

When the test material (BM) is, for example, a steel material that has been rolled and temporarily cooled, the surface of which the temperature is low during cooling becomes high due to recuperation. The temperature distribution in the width direction has a relatively large temperature variation, but the temperature change is gentle in the longitudinal direction. The defect portion is at a low temperature because the heat recovery after cooling is delayed, and there is a large temperature change there.

In the present invention, the main scanning direction x of the image pickup means (61) is set to a direction orthogonal to the moving direction of the test material (BM), that is, the width direction,
Since the sub-scanning direction z is parallel to the moving direction, the original image signal (O (z)) generated by the image pickup means (61) has a relatively large level variation on one main-scanning line, but the sub-scanning direction. Level fluctuation is small in the direction.

The smoothing means (321 to 326,328) is the original image signal (O (z))
Is smoothed in the sub-scanning direction to generate a smoothed image signal (P (z)). Therefore, this smoothed image signal (P (z)) is different from the original image signal (O (z)) in a defect-free area. Is small, so to speak, it can be called a predicted value. The difference between the original image signal (O (z)) and the smooth image signal (P (z)) is the deviation of the original state from the predicted value, and the large difference means that the original state greatly deviates from the prediction (non-defect surface). Means that When there is this deviation, the detection means (331
~ 333,301) detects that there is a defect, the defect detection accuracy is high. If the original image signal (O (z)) is smoothed in the main scanning direction (width direction of the steel material), the temperature change is large in the main scanning direction even on the surface having no defect. A region with a large deviation from the original image signal is generated, a region without a defect is erroneously detected as a defect, or the defective portion is overlooked.
Detection accuracy is low.

Furthermore, the present invention comprises a storage means (327) for holding a smoothed image signal (P (z)) for each pixel in the main scanning direction for one main scanning line, and the smoothing means (321 to 326, 328).
The original image signal (O (z)) of each pixel in the main scanning direction generated by the image pickup means (61) corresponds to the main scanning line immediately before the main scanning line to which it belongs and the corresponding pixel of the storage means (327). Of the smoothed image signal (P (z)) is smoothed in the sub-scanning direction to generate a smoothed image signal (P (z)), which is updated and stored in the storage means (327). , The difference (O (z)) between the original image signal (O (z)) and the smoothed image signal (P (z)) of the storage unit (327) up to the main scanning line immediately before the main scanning line to which it belongs. -P (z)) is calculated for the same position pixel in the main scanning direction and the difference image signal (O
Since (z) -P (z)) is generated, the difference image signal (O (z) -P (z)) is obtained simultaneously with the output of the original image signal (O (z)) from the image pickup means (61). Detection means (331-3
33, 301) are original image signals (O) output by the image pickup means (61).
It is possible to detect whether or not the pixel is a defective portion in synchronization with (z)), and the defect detection speed is high. As a result, it is possible to detect defects with high accuracy even in the case of a test material (BM) that moves at a relatively high speed.

In the embodiments described later of the present invention, the detection means (331 to 333,30
1) is based on the difference image signal (O (z) -P (z)) and is based on the smooth image signal (P
A count means (332) for counting the number of pixels in which the original image signal (O (z)) is darker than that of (z)), and a count data storage means (333) for storing the count value in correspondence with the sub-scanning position. Including. The count value is the total defect length (number of pixels; defect width) on one main scanning line, and this is stored in the count data storage means (333) in correspondence with the line sub-scanning position. Signal (O (z))
When one screen has been output, the count data storage means (33
In 3), the defect width data of each main scanning line of the one screen is present, and the defect information of the screen can be obtained substantially simultaneously with the image reading of the one screen. Fast defect information processing speed. As a result, even in the case of the test material (BM) that moves at a relatively high speed, the image reading can be repeated in a short cycle, and high-density defect detection can be performed.

Other objects and features of the present invention will become apparent from the following description of the embodiments with reference to the drawings.

〔Example〕

FIG. 1 schematically shows a defect detecting device for a steel material for carrying out the present invention as an example. This device is a system controller
1, input / output device 2, image processing units 3-1 to 3-4 (represented by "3" when it is not necessary to indicate each individually.
Same for everything else. ), Camera controllers 4-1 to 4-
4, motor controller 5-1 to 5-4, image pickup unit 6-
1 to 6-4, switching unit 7, monitor TV8, steel material detection sensor BSEN, transport speed sensor SSEN, and the like.

The imaging units 6-1 to 6-4 leave the heating furnace and rolling mill (not shown), and are cooled by water such as a sprinkler and a descaling device, and then an arrow is drawn by a steel material conveying device (not shown). The left and right flange surfaces and the upper and lower web surfaces of the steel material (here, H-shaped steel) BM conveyed in the direction at the line speed v are respectively imaged. Each imaging unit has its own IT
V camera 61, slit 62, case 63 accommodating ITV camera 61 and slit 62, rail 64 movably supporting ITV camera 61 and slit 62, and drive for independently changing the position of ITV camera 61 and slit 62 It consists of devices.

The case 63 is a heat-resistant dark box and has a window 63 with transparent glass.
1 and a reflecting mirror 632. The window 631 is the right side of the H-section steel BM in the unit 6-1 (toward the carrying direction).
To the flange surface on the left side of the unit 6-2, to the upper web surface of the unit 6-3, and to the lower web surface of the unit 6-4,
The openings are parallel to each other, and the reflection mirror 632 has a window 631.
The light incident via the LED is redirected to the ITV camera 61 without being twisted.

The ITV camera 61 is composed of a focal plane shutter and a light-receiving element array, and each light-receiving element outputs an electrical signal (hereinafter referred to as a density signal. “High density” corresponds to “high energy”) corresponding to the received light energy by 640 ×. Output in 480 pixel sections (in this, a raster scan is performed in which horizontal main scanning is performed and vertical sub-scanning is performed, and the sub-scanning direction is assumed to be parallel to the steel material conveying direction.).

The slit 62 corrects the deviation of the basic energy distribution of each surface of the steel material BM. For example, in the H-section steel, since the energy storage capacity at the boundary between the web and the flange is high, the recuperation after cooling becomes faster than that at the other parts. Therefore, the web surface has a temperature distribution as shown at the left end of FIGS. 2a and 2b, and the flange surface has a temperature distribution as shown at the left end of FIGS. 2c and 2d. Therefore, in the units 6-3 and 6-4 that capture the image of the web surface, the slits 62-3 and 62-4 attenuate the radiant energy from both ends of the web to receive the light receiving elements of the ITV cameras 61-3 and 61-4. In units 6-1, 6-2 that average the received light energy of the rows and image the flange surface, the slits 62-1 and 62-2 attenuate the radiant energy from the central part of the flange, and ITV camera 61-1 The received light energy of the light receiving element array 61-2 is averaged. These slits are driven by a driving device (not shown) and are positioned sufficiently close to the ITV camera so as to effectively average the received light energy of the light receiving element array.

In this embodiment, the slit used in the unit for imaging the web surface has a shape as shown in FIG. 3a.
A slit composed of a single slit plate and used for a unit for imaging the flange surface has a shape as shown in Fig. 3b.
Consists of a single slit plate (If the steel material to be measured is different, replace it if necessary. For example, when detecting flaws in a steel sheet pile, use two slit plates for the slit used in the unit that images the flange surface. There is also.) The positions of these slit plates are vertically changed by a driving device (not shown).

In response to a command from the system controller 1, the motor controller 5 controls the motor for changing the position of the ITV camera 61, the motor for changing the position of the slit 62, and the motor for changing the position of the slit plate of the slit 62, and controls them respectively. Position at the indicated position.

The camera controller 4 receives a command from the system controller 1 to control the shutter of the ITV camera 61, and drives the shutter at the instructed timing and the instructed speed.

The density signal output from the ITV camera 61 is branched into two in the switching unit 7, one is input to the image processing unit 3 and the other is selectively given to the monitor TV8.

The image processing unit 3 comprises a dedicated microcomputer 301 and various arithmetic circuits as shown in FIG. Here, for example, if the ITV camera 61 captures an image of a steel material surface having a flaw as shown in FIG. 5a,
Using it and its smooth image as shown in Figure 6a,
A difference image as shown in FIG. 7a is created and further binarized to obtain a binary image as shown in FIG. 8a. That is, when attention was paid to the sub-scanning line lv parallel to the conveying direction of the steel material passing through the flaw in the image and the density change in that direction was examined, as shown in the image shown in FIG. The data obtained by synthesizing the waveform corresponding to the temperature unevenness changing with the waveform corresponding to the flaw changing with the high frequency is obtained, and the temperature changing with the low frequency as shown in FIG. 6a from the image shown in FIG. 6a. Data corresponding to unevenness can be obtained. Therefore, from these difference images shown in FIG. 7a, data corresponding to only the flaws that change at high frequencies are obtained, and by binarizing the data, the flaw position in the sub-scanning direction as shown in FIG. 8b is obtained. Data that identifies

The smoothed data is the density data (IT
This is prediction data that predicts the density data of the pixel adjacent to the lower side by the primary delay element and the secondary delay element while sequentially capturing the data obtained by digitally converting the density signal of the V camera 61 by the A / D converter 311. . That is, the difference data is data that was not predicted from the concentration data that is continuous on the upper side. Specifically, the line buffers 327 and 328 for the number of main scanning pixels (640 pixels) are used to capture the density data of the pixels arranged on the sub-scanning line, and a constant (× α) is used.
The multiplication circuit 322 and the addition circuit 323 calculate a first-order lag element, and the addition circuit 324 calculates a second-order lag element (which originally includes a differential calculation, but here, the calculation is omitted in units of sampling cycle. , And a constant (× β) multiplication circuit 325 and an addition circuit 326 calculate the prediction data. The difference data is obtained by subtracting the predicted data from the density data of the pixel of interest (pixel of interest) at that time, and the subtraction circuit 321 performs the calculation.

When the density data of the pixel of interest is O (z) and the prediction data (smoothed data) at that time is P (z) (z indicates a sub-scanning address), the first-order lag element S (z) is , S (z) = P (z) + α [O (z) −P (z)] ...
(1) The second-order lag element S ′ (z) is S ′ (z) = S ′ (z−1) + [O (z) −P (z)].
(2) Prediction data P (z) of the next pixel adjacent to the lower side is given by P (z) = S (z) + β · S ′ (z) (3)

The difference data is compared with the binarization threshold value Sth given by the microcomputer 301 in the comparison circuit 331, and is “1” when the density is higher (lighter) and lower (darker).
It is sometimes binarized as "0". This binarized data is input to the counter 332.

The counter 332 is reset by the sub-scanning synchronization signal and counts up at the trailing edge of binary data “0”. This count data is input to the shift register 333 at the end of each main scan. Therefore, when the processing of one frame is completed, the shift register 333 stores 1 "0" pixel of the binary image.
The data projected on the sub-scanning line of the book can be obtained. That is, for example, if a binary image including flaws 1 and 2 given by “0” as shown in FIG. 10a is obtained, the shift register 333 has a binary image as shown in FIG. 10b. Then, vertical projection data F (z) corresponding to the histogram showing the appearance degree of "0" pixel in the sub-scanning direction is obtained (however, each vertical projection data is specified by the sub-scanning address z). In the microcomputer 301, this vertical projection data F
(Z) is compared with a predetermined threshold value Wth, and the data group continuously exceeds the threshold value Wth in the sub-scanning direction (corresponding to the binary data of FIG. 10c obtained by binarizing the histogram of FIG. 10b).
After extracting the feature quantity with the maximum number of pixels in the main scanning direction as "width", the number of pixels in the sub-scanning direction as "height", and the central sub-scanning address as "position", Each defect is detected by examining based on the defect judgment standard given by the system computer 1.

In the image processing unit 3, the vertical projection data F
The average density data M is generated during the detection of (z). This average density data M is calculated as IT as shown in FIG. 9a.
V Specify the specified value P above and below the center of the screen of the camera 61
Of the pixels on the main scanning lines l ₁ and l ₂ that are separated by / 2, the density data is averaged for those whose density data exceeds a predetermined value Δ, and is used for adjusting the shutter speed of the ITV camera 61. Here, δ is the line l _{1 as} shown in FIG. 9b when the shutter speed is appropriately set.
Or it is the value that divides the concentration distribution on l ₂ into the background part and the steel part.

The average density data M is calculated by the gate circuit 341, the addition / subtraction circuit 342, the buffer 343, the comparison circuit 344, the counter 345 and the microcomputer 301 shown in FIG. Comparison circuit 3
44 compares the density data with a predetermined value δ to generate a gate signal for permitting “gate opening” when the density data exceeds δ, and supplies the gate signal to the gate circuit 341. The gate circuit 341 opens the gate and sends the concentration data to the addition / subtraction circuit 342 side only when the "gate opening" is permitted by this gate signal. The addition / subtraction circuit 342 and the buffer 343 accumulate the difference between the sent density data and the temporary average density data D ₀ to generate difference accumulated data D. During this period, the counter 345 counts the number of pixels for which "gate opening" is permitted, and generates count data N. The temporary average density data D ₀ is used to reduce the number of bits of the adder / subtractor circuit 342,
The buffer 343 and the counter 345 are reset by the sub scanning synchronizing signal.

In the microcomputer 301, the difference accumulation data D and the count data N are read at the end of the main scanning of the l ₁ line and the l ₂ line, the average D / N of the difference accumulation data D is calculated, and the provisional average density data D _{0 is obtained.} Average density data M
To generate.

Next, the overall operation of the apparatus of this embodiment will be described. See FIG. 11.

When the operator inputs the product type code and rolling No. of the steel to be measured through the input / output device 2, the system controller 1 refers to the data stored in advance to perform IT.
Imaging conditions related to the position of the V camera 61, the shutter speed (initial value), the position of the slit 62, the position of the slit plate, the binarization threshold Sth, and the flaw determination standard (W1 _L , W2 _L , H1 _L , H
2 _L ) and other image processing conditions are set and transferred to the motor controller 5, the camera controller 4 or the image processing unit 3, respectively, and the standby mode is set. As a result, the motor controller 5 positions the ITV camera 61, the slit 62, and the slit plate at the designated position, and the camera controller 4 sets the designated shutter speed. Further, the microcomputer 301 of the image processing unit 3 initializes the input / output port, the internal register, etc., then registers the image processing condition, and gives the binarization threshold value Sth to the comparison circuit 331 to set the standby mode.

After that, when the steel material detection sensor BSEN detects the tip of the steel material to be measured, the system controller 1 and the microcomputer 301 of the image processing unit 3 start up, and the system controller 1 responds to the steel material conveyance speed detected by the speed sensor SSEN. The imaging interval (interval for performing image processing) is set. This imaging interval is set to 15 Hz when the steel material conveying speed exceeds 4 m / s, and 10 Hz when it is 4 m / s or less. In other words, the ITV camera 61 captures one screen in 1/30 second, so 15H
When the imaging interval of z is set, as shown in Fig. 12a, 1
If you set a 10Hz imaging interval on the screen, the 12b
As shown in the figure, image processing is performed every two screens.

Further, the system controller 1 calculates the effective area according to the transport speed of the steel material. This effective area limits the flaw detection range, and its length in the sub-scanning direction.
Va is Va = v · fs, where v is the transport speed of the detected steel material, v ′ is the shutter speed of the focal plane shutter, fs is the imaging cycle, Cpm is the length of one pixel, and ε ₀ is the overlap rate. ^-1・ Cmp ^-1・ v '/ (v + v') ・ ε
It is given by ₀ (4). The effective area is set for each sampling screen (screen for image processing) as shown in FIGS. 12a and 12b, and the effective areas before and after overlap with the overlap ratio ε ₀ , so Defects are included in any of the effective areas. When the system controller 1 sets the effective area based on the above equation (4), the frame N
The length Va is registered in association with the o. Instruct to start processing.

Upon receiving an instruction to start image processing, the microcomputer 301 performs image processing on the screen imaged by the ITV camera 61 at that time. FIG. 13 is a flowchart showing the average density data detection subroutine performed at the beginning of image processing. This subroutine is activated at the end of the main scan of line l ₁ or l ₂ .

As described above, the difference accumulation data D is provided from the output end of the buffer 343 and the count data N is provided from the output end of the counter 345 at the end of each main scan. Therefore, the microcomputer 301 reads the difference accumulation data D and the count data N respectively at the end of the main scanning of the lines l ₁ and l ₂ , calculates D / N + D _0, and obtains the average density data of each line. When the average density data of each line is averaged to obtain the average density data M of the screen, the average density data M is transferred to the system controller 1.

When the average density data M is less than or equal to the predetermined value, the system controller 1 determines that the steel material has not reached the image capturing position (the steel material detecting sensor BSEN is installed upstream from the image capturing position). I will wait for the image of the screen of
When it exceeds a predetermined value, it is determined whether the shutter speed is appropriate. In determining the appropriateness of the shutter speed, the average density data M and preset reference levels D ₁ , D ₂ , U ₁ , U
₂ (however, D ₂ <D ₁ <U ₁ <U ₂ ) is compared, and when the average density data M is the reference level D ₂ or less, it exceeds “−2”, D ₂ and D ₁
In the following cases, the evaluation value is “−1”, when it is over D ₁ and U ₁ it is “0”, when it is over U ₁ and U ₂ it is “+1”, and when it is over U ₂ it is “+2”. Set. Here, the sign of the evaluation value indicates excess or deficiency of exposure (minus corresponds to insufficient, plus corresponds to excessive), and the number indicates the degree of excess or deficiency. Therefore, for example, when the evaluation value "-2" is set, since the exposure is largely insufficient, the camera controller 4 is instructed to reduce the shutter speed by two steps (slow) and the evaluation value "+1". When set to, the exposure is a little too large, so
The camera controller 4 is instructed to increase the shutter speed by one step (speed up).

After that, the system controller 1 instructs the microcomputer 301 of the image processing unit 3 to execute the flaw detection process.

14a and 14b are flowcharts showing a subroutine of flaw detection processing. This subroutine is instructed by the system controller 1 to execute a flaw detection process, and is activated at the end of screen collection. Hereinafter, this process will be described step by step.

As described above, at the end of screen collection, the counter 333
Since the vertical projection data F (z) is output from the output end of
First, in step 2, this data is read, and in step 3, the registers H, W, G, Z, i and z are cleared (0).
To do.

The register G is a flag for detecting a group of vertical projection data F (z) that continuously exceeds the threshold value Wth in the sub-scanning direction, and if this value is 0, the register z (corresponds to the sub-scanning address). Threshold value Wth
Search for vertical projection data F (z) that exceeds (step 4,
5,13,14). During this period, the vertical projection data F that exceeds the threshold value Wth
When (z) is found, the value of the register G is set to 1 in the step 6 and the register i is incremented by 1,
In step 7, the value of the register W (i) specified by the value of the register i (hereinafter, referred to as the register W (i). The same applies to the rest) (initially 0) is compared with the data F (z). At this time, the value of the register W (i) is the data F (z).
If smaller, register W (i) in step 8
The data F (z) is stored in, the value of the register H (i) is incremented by 1 in step 9, the value of the register z is incremented by 1 in step 13, and step 4
Return to.

This time, since the value of the register G is 1, the vertical projection data F (z) is compared with the threshold value Wth in step 10, and steps 10 and 7 are performed until the vertical projection data F (z) becomes equal to or less than the threshold value Wth. Repeat ~ 9, 13, 14 and 4. After that, when the vertical projection data F (z) becomes less than or equal to the threshold value Wth, step 1
The register G is cleared in 1 and the value obtained by subtracting 1/2 of the value in the register H (i) from the value in the register z is stored in the register Z (i) in step 12. That is, at this point, of the groups of the vertical projection data F (z) exceeding the threshold value Wth, the “width (maximum number of pixels in group)” of the i-th group from the top is stored in the register W (i) as “high”. "(The number of pixels in the sub-scanning direction of the group)" is stored in the register H (i),
The "position (sub-scanning address of the central pixel of the group)" is stored in the register Z (i).

The above processing is performed for all the vertical projection data F (z), and the feature amount of the group of the vertical projection data F (z) exceeding the threshold value Wth is registered in the registers W (i), H (i) and Z (i).
Then, each group is examined by comparing the feature amount with the defect judgment standard.

Since the number of groups detected at this time is indicated by the value of register i, the value of register i is saved in register I in step 15, and registers Aw, Ah, Bw and Bh are cleared (0) in step 16 Store l in i, j and k.

As described above, since the range from the sub-scanning address Vs to the sub-scanning address Ve is set as the effective area by the system controller 1, first, the value of the register i is updated and stored in the register Z (i). A group of vertical projection data F (z) whose position data is in this range is searched (steps 17, 26, 27).

Vertical projection data F whose position data is included in the effective area
When the group of (z) is found, the width data of the group stored in the register W (i) in step 18 is compared with the first width determination reference W1 _L, and stored in the register H (i) in step 19. The height data of the group being compared with the first height criterion H1 _L are compared. In these comparisons, when the width data exceeds the first width determination criterion W1 _L or the height data exceeds the first height determination criterion H1 _L , the group of the vertical projection data F (z) is class A.
It is determined that the value of the register W (i) is stored in the register Aw (j) and the value of the register Ah is stored in the register Ah (j) in step 20.
The value of register H (i) is stored in (j), and the value of register j is incremented by 1 in step 21.

Also, the width data of the group being examined at this time is the first
When the width is less than the width criterion W1 _L and the height data is the first height criterion H1 _L , the width data of the group and the second width criterion are further determined in steps 22 and 23.
Comparison with W2 _L and height data with the second height criterion H2 _L (W1 _L > W2 _L , H1 _L > H2 _L ) are performed. In this, the width data exceeds the second width criterion W2 _L , and
When the height data exceeds the second height criterion H2 _L ,
It is determined that the group of the vertical projection data F (z) corresponds to the flaw of class B, and the register Bw is registered in step 24.
The value of the register W (i) is stored in (k), the value of the register H (i) is stored in the register Bh (k), and the value of the register k is incremented by 1 in step 25.

The above examination is performed for all groups of the vertical projection data F (z), and the detected feature quantities (width and height) of the class A flaw are registered in the registers Aw and Ah, and the feature quantity of the class B flaw (width and height). When the height) is sorted in the registers Bw and Bh, the data is stored (registered) in step 28 in association with the frame number.

The above processing is repeated for each set imaging cycle until the steel material detection sensor BSEN detects the absence of steel material and the average density data M becomes a predetermined value or less (no steel material at the imaging position).

The system controller 1 notifies the microcomputer 301 of the image processing unit 3 of the end of processing when the average density data M becomes equal to or less than a predetermined value after the steel material detection sensor BSEN detects that there is no steel material. This allows the microcomputer
Since the data related to the flaw registered in association with the frame number is transferred from 301, the system controller 1 uses the effective area length Va stored in association with the frame number to store the data. The defects are sorted and displayed in the CRT display of the input / output device 2 in association with the distance from the front end and the distance from the rear end of the steel material, the class thereof, and the trace amount, and are also printed out via the printer.

In the above embodiment, the deviation of the basic energy distribution on the imaging surface of the steel material is corrected by the slit, but this may be done by masking the output signal of the ITV camera 61. However, since the temperature difference that occurs at the flaw is small over the entire range of energy on the imaging surface, it is necessary to use IT when correcting the bias in the energy distribution by signal processing.
Care must be taken when selecting a V camera.

Further, it is obvious that the present invention can be applied to the flaw detection of various materials other than the H-section steel shown in the above embodiment.

〔The invention's effect〕

As described above, according to the present invention, when the steel material that has been rolled and temporarily cooled is the test material (BM), the smooth image signal (P (z)) shows the original image in the defect-free region. The deviation from the signal (O (z)) is small, so to speak, it can be called a predicted value. The difference between the original image signal (O (z)) and the smooth image signal (P (z)) is the deviation of the original state from the predicted value, and the large difference means that the original state greatly deviates from the prediction (non-defect surface). Means that When there is this deviation, the detection means (331 to 333, 301) detect that there is a defect, so the defect detection accuracy is high. If the original image signal (O (z)) is smoothed in the main scanning direction (width direction of the steel material), the temperature change is large in the main scanning direction even on the surface having no defect. The detection accuracy is low, for example, a region having a large deviation from the original image signal is generated, a region having no defect is erroneously detected as a defect, or a defective portion is overlooked.

Furthermore, in the present invention, the original image signal (O
(Z)) and the difference image signal (O (z) −
P (z)) is obtained, and the detection means (331 to 333, 301) detect whether or not the pixel is a defective portion in synchronization with the original image signal (O (z)) output by the image pickup means (61). The defect detection speed is high. As a result, it is possible to detect defects with high accuracy even in the case of a test material (BM) that moves at a relatively high speed.

In the above-described embodiment of the present invention, the defect information of the screen can be obtained substantially simultaneously with the image reading of the screen. Fast defect information processing speed. As a result, even in the case of the test material (BM) that moves at a relatively high speed, the image reading can be repeated in a short cycle, and high-density defect detection can be performed.

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a schematic diagram showing the configuration of a steel defect detecting apparatus embodying the present invention as an example. 2a to 2d are explanatory views for explaining the correction of the deviation of the energy distribution of the H-section steel. 3a and 3b show the slit 62 used in the embodiment apparatus.
It is a top view which shows the concrete shape of. FIG. 4 is a block diagram showing a detailed configuration of the image processing unit 3 provided in the apparatus of the embodiment. Figures 5a, 5b, 6a, 6b, 7a, 7b, and
FIG. 8a and FIG. 8b are explanatory diagrams for explaining the outline of the flaw detection processing performed in the image processing unit 3. FIG. 9a and FIG. 9b are explanatory diagrams for explaining the average density data detection processing performed by the image processing unit 3. 10a to 10c are explanatory views for explaining extraction of flaw feature data performed by the image processing unit 3. FIG. 11 is a flowchart showing the overall operation of the apparatus of the embodiment. 12a and 12b are explanatory views for specifically explaining the setting of the imaging interval. FIG. 13 is a flowchart showing the average density data detection processing performed by the image processing unit 3. FIGS. 14a and 14b are flowcharts showing the flaw detection processing performed by the image processing unit 3. 1 ... System controller 2 ... Input / output device 3 ... Image processing unit (smoothing means, processing means, detection means) 4 ... Camera controller 5 ... Motor controller 6 ... Imaging unit (imaging means) 7 ... Switching Unit 8: Monitor TV

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Munekazu Hirano 1 Kimitsu, Kimitsu-shi, Chiba Nippon Steel Stock Co., Ltd. Inside the Kimitsu Works (56) Reference JP-A-49-40658 (JP, A) Sho 63-19078 (JP, A)

Claims (2)

[Claims] 1. A main scanning direction x is defined as a direction orthogonal to the moving direction, in which an image of the surface of a test material moving in a predetermined direction is imaged and an original image signal whose optical characteristics are converted into an electric signal is generated. Image pickup means in which the sub-scanning direction z is parallel to the moving direction; storage means for holding a smoothed image signal for each pixel in the main scanning direction for one line of main scanning; main scanning direction generated by the image pickup means The original image signal of each pixel is smoothed in the sub-scanning direction by using the smoothed image signal of the corresponding pixel in the storage means up to the main scanning line immediately before the main scanning line to which it belongs, and a smoothed image signal is generated. Smoothing means for updating and storing in the storage means: Difference between the original image signal and the smoothed image signal in the storage means up to the main scanning line immediately before the main scanning line to which the original image signal belongs corresponds to the same position pixel in the main scanning direction Give a signal Processing means for; and, detecting means for detecting a defect in the surface of the test material on the basis of the differential image signal the processing unit has generated; surface defect detection apparatus comprising a. 2. The detecting means counts the number of pixels in each main scanning line where the original image signal is darker than the smoothed image signal based on the differential image signal, and the count value corresponds to the sub-scanning position. The surface defect detection device according to claim (1), further comprising: count data storage means for storing the count data.