JP3015325B2 - Streak inspection method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a stripe inspection method and a stripe inspection apparatus for easily detecting stripes having a low contrast, a narrow width, and a long stripe in a transport direction.

[0002]

2. Description of the Related Art For example, in a production line for processing sheet-like objects such as paper, film, and steel plate, scratches, irregularities,
It is necessary to inspect for defects such as dirt and foreign matter. In order to inspect such a defect, a line sensor camera is generally provided on a conveyed sheet to image the sheet. Then, image processing is performed on the video signal captured by the line sensor camera to emphasize defects.

In this image processing, minute defects are detected by micro-filter processing on image data obtained by digitally converting a video signal from a line sensor camera (see Japanese Patent Application Laid-Open No. 7-1995).
225196), and dark unevenness such as a pattern or dark unevenness such as a dirt is detected by an unevenness filter process (see Japanese Patent Application Laid-Open No. 6-323954). That is, the inspection is performed using a different inspection method according to the type of the defect.

[0004]

However, the microfiltering process emphasizes the amount of change in density information between pixels and averages the amount of change.

[0005] Then, when the difference between the obtained averaged amounts of change is equal to or greater than a predetermined threshold value, it is detected as a fine defect.

In other words, the determination by the micro-filter processing emphasizes the difference between the density values of one pixel, and when the difference is equal to or more than a predetermined value, it is determined that the pixel is defective.

For example, a streak having a low contrast, a narrow width, and a long length in the transport direction cannot be detected because the density value is so small that the difference becomes so small that the difference does not exceed the threshold value.

Therefore, there is a problem that a defect with low contrast cannot be detected by the microfilter.

[0009] The non-uniformity filter process divides image data, which is density information of a pixel, into a grid composed of a predetermined number of pixel matrices each in the vertical and horizontal directions, and integrates the density information of each pixel in each grid. The unevenness is emphasized, and the difference between the integral values is obtained, so that when the obtained difference amount is equal to or more than a predetermined value, the unevenness is determined.

That is, since the unevenness filter process integrates the density values of a plurality of pixels and regards the difference as unevenness, there is a problem that unevenness is not detected unless there is a certain area.

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and an object of the present invention is to provide a method and an apparatus for detecting a stripe which can easily detect a stripe having a low contrast, a narrow width, and a long stripe in a transport direction. And

[0012]

SUMMARY OF THE INVENTION In accordance with the present invention, there is provided a method for inspecting a stripe, which scans a sheet-like inspection object while transporting a sheet-like inspection object, and converts an imaging signal resulting from the imaging to each pixel. In an inspection method for sending image data to each subsequent step and inspecting for the presence or absence of a linear stripe formed on the inspection object, input image data for each pixel obtained by converting an imaging signal, A change amount calculating step of emphasizing the amount of density change in the vertical and horizontal directions by a first emphasis filter of m rows and n columns with respect to the image data; Cut out with a grid of data,
An average calculation step of setting an average value of the plurality of cut-out data as an average change amount, and cutting out the change amount averaged by the average calculation step into a grid including a plurality of vertical and horizontal matrices, and adding the change amounts in this grid. A change amount adding step for obtaining an added value for each lattice; and p for the change amount obtained in the change amount adding step.
The second emphasis filter in the row q column detects a streak by comparing a change amount calculating step of emphasizing the amount of density change in the vertical and horizontal directions and the change amount emphasized in the change amount calculating step with a predetermined threshold value. A defect determination step, a change amount calculation step,
The gist is that serial processing is performed in the order of the average calculation step, the change amount addition step, the change amount calculation step, and the defect determination step.

A streak inspection apparatus according to the present invention comprises: a camera for imaging a test object while conveying a sheet-like test object; and a post-stage image data for each pixel obtained by converting an image signal accompanying the photographing. , An image input unit to be sent to the processing unit, and image data sent from the image input unit, and a density change in the vertical and horizontal directions is applied to the image data by a first emphasis filter of m rows and n columns. A change amount calculating means for emphasizing an amount, and a density change amount emphasized by the first change amount calculating means is cut out by a grid composed of a plurality of data, and an average value is calculated using an average value of the plurality of cut out data as an average change amount. A micro filter unit serving as a means, and a change amount processed by the micro filter unit is cut out into a grid including a plurality of vertical and horizontal matrices,
An uneven sector adder for obtaining an added value for each grid by adding the amount of change in the grid, and a vertical / horizontal filter for a p-row and q-column second enhancement filter for the added value obtained by the uneven sector adder. A variation calculating unit for enhancing the density variation in the direction; and a defect determining unit for comparing the variation emphasized by the variation filtering unit with a predetermined threshold to detect a streak and determine a defect. And the image input section,
The gist is that the micro filter section, the uneven filter section, and the defect determination section are configured in series in this order.

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 is a schematic configuration diagram of a stripe inspection apparatus according to the present embodiment. Streak inspection device 1 of FIG.
A line sensor camera 3 for photographing the sheet 2, an illuminator 5 provided near an imaging line for imaging by the line sensor camera 3, and irradiating light based on electric power from an illumination power supply 4; A sensor 7 provided at a position upstream of the camera 3 for detecting the sheet 2; and when the sensor 7 detects a sheet, a video signal from the line sensor camera 3 is fetched every predetermined line, and an image of the fetched line is taken. An inspection control device 11 is provided for detecting streaks (including low-contrast streaks) from data and outputting the detection result to the personal computer 9 and the printer 10.

Further, the inspection control device 11 is provided with a monitor 12 for displaying an image of the sheet 2.

As shown in FIG. 1, the inspection control device 11 of the stripe inspection device 1 includes an encoder 21 and
Image input unit 22, frame memory 24, micro filter unit 25, unevenness filter unit 26, defect determination unit 27
And The above-described encoder 21 outputs pulse signals at fixed intervals.

When the sensor 7 detects the sheet 2, the encoder 21 outputs a pulse signal having a width corresponding to the image separation width to the image input unit 22.

The image input unit 22 synchronizes the video signal for each line from the line sensor camera 3 with the pulse signal from the encoder 21 in accordance with the input of the pulse signal from the encoder 21, and Outputs digital image data (pixel density).

The micro filter unit 25 includes an image input unit 22
, The amount of change in density information between pixels is emphasized, and the amount of change is averaged. As a result, data emphasizing a defect (contour such as minute unevenness or pinhole) is output.

The unevenness filter section 26 divides the image data, which is the density information of the pixels, into a grid composed of a predetermined number of pixel matrices in each of the vertical and horizontal directions, and integrates the density information of each pixel in each grid to emphasize unevenness. are doing.

The defect judging unit 27 judges that the unevenness emphasized by the unevenness filter unit 26 has a predetermined length as a streak.

The defect determination section 27 outputs the inspection result to the personal computer 9 or outputs the inspection result to the printer 10.
Output to

Next, FIG. 2 shows a state in which an image of the sheet 2 shown in FIG. That is, the sheet 2 is being conveyed in the direction of the arrow. Scratch 2
8 is directed to a thin and long linear scratch or defect parallel to the transport direction as shown in the figure. Particularly in the present invention,
Even when the contrast ratio between the stripe and the surrounding area is low, the stripe can be detected satisfactorily. The line A-A in FIG. 2 indicates a location that is imaged by the line sensor camera 3.

FIG. 3 (a) shows the stripe scratches 28 shown in FIG.
Indicates an image signal of one line captured on the line AA using the line sensor camera 3, and is input to the inspection control device 11. That is, FIG. 3A shows the sheet width on the horizontal axis and the brightness on the vertical axis, and a signal representing the stripe 28 is recognized above this signal. However, the signal representing the streak 28 has a smaller difference in brightness and difference in width, that is, a difference in contrast as compared with a signal of a portion having no flaws in the surroundings. Even so, its detection was difficult.

FIG. 3B schematically shows a data structure in which the image signals shown in FIG. 3A are transferred to a frame memory 24 (described later) while the image pickup line is shifted while the sheet is conveyed. It is. That is, although FIG. 3B has a lattice shape, each lattice corresponds to one pixel. The horizontal direction in the drawing indicates the width direction of the sheet 2, and the vertical direction in the drawing indicates the conveying direction of the sheet 2. The data described in each grid indicates the degree of lightness or darkness. The darkest state is represented by 0, and the brightest state is represented by 255. FIG. 3B is directed to the stripe scratch 28 in FIG.
This is a data structure using the image signal shown in FIG. 3A as an example. That is, in FIG. 3B, data 145 is stored linearly in the column of the stripe 28, and data 150 and the like are stored in the other columns. This streak scratch 28
Column corresponds to the brightness of the streak 28, and its periphery represents a regular region. Thus streak scratches 28
Is very low, that is, a low contrast state.

Next, details of the inspection control device 11 will be described with reference to FIG. The image input unit 22 includes an A / D converter, a shading correction unit, a projection calculation unit, a luminance correction unit, and the like.

When the sheet 2 is detected by the sensor 7 and a pulse signal is input from the pulse controller 21, the image input section 22 synchronizes the video signal from the line sensor camera 3 with the pulse signal to generate a digital signal. After that, image data that has been subjected to shading correction, projection correction, and luminance correction is transmitted.

As shown in FIG. 4, the microfilter unit 25 is composed of a vertical sobel filter 31 of three rows and three columns, which is a spatial filter for performing a change amount operation, and an average value filter 32 for calculating an average value. I have.

The microfilter unit 25 calculates the amount of change in the density in the vertical and horizontal directions of the nine grid regions in three rows and three columns using the vertical sobel filter 31, and calculates the amount of change in each pixel. Each time, an average value of the added change amounts is obtained using the average value filter 32. The average is 3
By taking the average of the amounts of change of the center pixels of the nine grid regions in the row 3 column, the horizontal direction of the stripe 28, that is, the sheet 2
Can be emphasized in the width direction.

As shown in FIG. 5, the vertical Sobel filter 31 includes -1, -2,-
1 is set to 0, 0, 0 in the middle column, and 1, 2, 1 from the top in the right column.

For example, the horizontal change amount Hij of an arbitrary pixel Lij is calculated based on the density information of nine grids of three rows and three columns centered on the pixel Lij as shown in FIG. The Sobel filter 31 is obtained by the following equation (1).

[0040]

H _ij = | −L _{j−1, j−1} + L _{j + 1, j−1} −2L _{i−1, j} + 2L _{i + 1, j} −L _{i−1, j + 1} + l _{i + 1, j + 1} |... Equation (1) The change amount Hij in the horizontal direction is obtained by the above equation (1).

The calculation using the vertical sobel filter 31 as in the above equation (1) will be described by taking the data structure shown in FIG. 3B as an example. That is, the product-sum operation processing is performed using the vertical sobel filter 31 on the three rows and three columns indicated by the thick frame in FIG. The calculation shown by the thick frame in FIG. 7 is performed by shifting one image from left to right one image at a time, and then is performed by shifting down one stage and performing the calculation again from left to right. FIG. 7B shows one row 33 in FIG.
In FIG.

FIG. 8 shows the result of the product-sum operation performed on the nine grids, that is, for each of the nine pixels. FIG.
One grid in FIG. 7A shows the result of the product-sum operation performed on the nine pixels in FIG. 7A using a vertical Sobel filter. In this manner, it can be seen that the data having a low contrast difference in the density value of the pixel unit in FIG. 7 has the contrast difference enhanced as shown in FIG. 8A. FIG.
FIG. 8B is a diagram illustrating data of one row 34 in FIG. 8A as an image signal. That is, a signal corresponding to -20 in FIG. 8A is represented by a negative peak 34a, and a signal corresponding to +20 is a positive peak 34a.
It is represented by b.

FIG. 9 shows the absolute values of the data shown in FIG. 8, FIG. 9A shows the numerical values, and FIG. 9B shows the values as image signals. In this manner, as the low-contrast streak data shown in FIG. 7, data in which the feature is emphasized as shown in FIG. 9 is extracted.

The image data shown in FIG.
Are averaged by using the average value filter 32. This averaging is performed by obtaining an average value for the data of FIG. 9A for every three rows and three columns.

FIG. 10 shows an image signal before performing the averaging process shown in FIG. 9B. The height of the signal corresponding to the stripe is S, the width is W, and the signal around the stripe is shown. The height is represented as N. FIG. 11 shows an image signal after the averaging process, in which the height of a signal corresponding to a stripe is denoted by s, the width is denoted by w, and the height of a signal around the stripe is denoted by n. In this way, the feature of the stripe is further extracted, and n / N <s / S, and further, W / w. As a result, the width of the signal corresponding to the streak is expanded as compared with before the averaging. However, even in this state, accurate and sufficient extraction of stripe marks cannot be performed. The effect of the present invention can be obtained by the unevenness filter processing.

The data thus averaged is further processed by the unevenness filter unit shown in FIG. The averaged data is stored in each of the grids 36 as shown in FIG. The mura-sector adder 35 calculates an addition value の for each 6 × 30 grid, and this addition is sequentially shifted. As a result, grid data of the added value よ う な as shown in FIG. 13B is obtained.

As described above, in order to extract the characteristics of a thin and long linear streak, the unevenness adder 35 uses as many as 30 grid data in the streak direction and reduces six grid data in a direction orthogonal to the streak. Using. That is, FIG.
Addition is performed in units of vertically long regions as indicated by a thick frame 3.

FIG. 14 shows, for example, a row 37 as an image signal in the lattice data of the added value sigma shown in FIG.
It is. As a result, compared to a gentle signal such as the signal in FIG. 13, a signal with good sharpness in which the boundaries between the signals are vertical as shown in FIG. 14 is obtained.

Thereafter, the vertical sobel filter 38 shown in FIG.
Is used to calculate the amount of change. This vertical sobel filter 3
Reference numeral 8 is similar to the vertical sobel filter 31 shown in FIG.

For example, as shown in FIG. 13 (b), the amount of change Sij in the horizontal direction of the density addition value 、 ij of an arbitrary grid is determined by the density addition of nine grids in three rows and three columns centered on the density addition value Σij. Based on the values, the vertical sobel filter 3 shown in FIG.
1 and the following equation (2).

[0051]

S _ij = | Σ _{i-1, j + 1} + Σ _{i + 1, j-1} -2Σ _{i-1, j} + 2Σ _{i + 1, j} −Σ _{i-1, j + 1} + Σ _{i + 1 , j + 1} | ... Equation (2) As described above, the density information for each pixel of the inspection object obtained by the line sensor camera 2 is divided into a grid composed of a plurality of vertical and horizontal pixel matrices. The unevenness is emphasized by adding the density information to obtain the density addition value for each grid, and the change amount Si of the horizontal density addition value in each horizontal direction between the grids.
The unevenness is further emphasized by obtaining j in a matrix of 3 rows and 3 columns.

That is, the defect data after the micro-filter processing is arranged on a lattice and integrated by a vertical and horizontal Sobel filter along the arrangement direction to obtain uneven defect data based on the directionality of the stripe.

Therefore, the 3 × shown in the bold frame in FIG.
The product-sum operation is performed on the area No. 3 using the vertical Sobel filter 38. As a result, a signal as shown in FIG. 15 is obtained in which the stripes are emphasized and the boundary with the surroundings is clear.

(Explanation of the Overall Operation) The operation of the stripe inspection apparatus configured as described above will be described with reference to FIG.

For example, when the sheet 2 having the low-contrast streaks 28 is conveyed and this sheet is detected by the sensor 7, a detection signal is output from the sensor 7 to the inspection control device 11 (step ST1). .

In accordance with the input of the detection signal, a one-line video signal from the line sensor camera 3 is transmitted to the image input unit 2.
2 is input and digitally converted to obtain image data corrected based on an optical error (step ST2).

This image data is input to the micro filter unit 25, and the vertical Sobel filter 31 of the change amount calculation means of the micro filter unit 25 sets the vertical direction Hij between the density information for the pixel density information in three lines. It is determined within a matrix of three columns (step ST3).

Next, the average value filter 33 of the average calculating means is used.
Sets the average of the nine changes in three rows and three columns to the average of the center pixel, and sets the average of the obtained changes to the average change (step ST4).

Accordingly, the minute streaks having a low contrast are enlarged by several bits by the micro filter unit 25, and are output in a defect size that can be detected by the uneven filter unit 26.

The average change amount obtained by the micro filter unit is input to the unevenness filter unit 26 and the following processing is performed.

The unevenness filter unit 26 receives the micro defect data from the microfilter unit 25 by the unevenness sector adder 35, and the sector adder 35 divides the density information into a grid consisting of a plurality of vertical and horizontal pixel matrices (step ST5). The density information of each pixel in each pixel in each grid is added to obtain a density addition value for each grid (step ST6).

Next, the change amount calculating means 38 of the unevenness filter section 26 obtains the vertical change amount Sij between each lattice in a matrix of 3 rows and 3 columns (step ST7).

Then, the defect determination section 27 determines the vertical or horizontal length of the obtained unevenness defect data, and if the length is longer than a predetermined length, determines that there is a stripe in that direction (step ST8). .

Therefore, even if a low-contrast fine stripe is detected by the line sensor camera 2, the low-contrast stripe is enhanced by the serially connected micro-filter unit 25 and uneven filter unit 26, and the defect is judged by the defect determination unit 27. It is output, and if it is longer than a predetermined length, it is determined as a stripe.

[0065]

As described above, according to the present invention, each time a sheet being conveyed is imaged, the obtained image data is emphasized by the micro filter unit using a spatial filter to enhance the contrast difference. That is, the defect size is set to a size that can be detected by the unevenness filter unit. Then, a non-uniformity filter unit provided immediately after the micro filter unit sets a sufficiently large predetermined area in the vertical direction with respect to the image data after the contour enhancement in the horizontal direction, and adds and averages each pixel density information. Do.
That is, unevenness based on the direction of the stripe is emphasized. Therefore, an effect is obtained that even a fine defect having a low contrast can be easily determined as a streak.

[0066]

[0067]

[0068]

[0069]

[0070]

[0071]

[0072]

[0073]

[0074]

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a line mark inspection apparatus according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram for explaining a streak of a sheet.

FIG. 3 is an explanatory diagram for explaining a streak of a sheet.

FIG. 4 is a schematic configuration diagram of a micro filter unit.

FIG. 5 is an explanatory diagram illustrating a vertical sobel.

FIG. 6 is an explanatory diagram illustrating calculation of a change amount of a micro filter unit.

FIG. 7 is an explanatory diagram illustrating a product-sum operation process of a microfilter unit.

FIG. 8 is an explanatory diagram illustrating a result of the product-sum operation processing of the microfilter unit.

FIG. 9 is an explanatory diagram illustrating a change amount calculation result of the micro filter unit.

FIG. 10 is an explanatory diagram illustrating an averaging process of a micro filter unit.

FIG. 11 is an explanatory diagram illustrating a result of an averaging process performed by a microfilter unit.

FIG. 12 is a schematic configuration diagram of an unevenness filter unit.

FIG. 13 is an explanatory diagram illustrating calculation of a change amount of the unevenness filter unit.

FIG. 14 is an explanatory diagram illustrating calculation of an addition value of the unevenness filter unit.

FIG. 15 is an explanatory diagram for calculating a change amount of an addition value of the unevenness filter unit.

FIG. 16 is an explanatory diagram illustrating a flow of a streak inspection according to the present invention.

[Explanation of symbols]

 2 sheet 3 line sensor 4 lighting power supply 5 illuminator 7 sensor 11 inspection control device 21 encoder 22 image input unit 24 frame memory 25 micro filter unit 26 uneven filter unit 27 defect determination unit 28 stripe marks

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int. Cl. ⁷ , DB name) G06T 1/00, 7/00 G01B 11/30 G01N 21/88

Claims (4)

(57) [Claims] 1. An image of the inspection object is taken while a sheet-like inspection object is being conveyed, and image data for each pixel obtained by converting an imaging signal accompanying the imaging is sent to each subsequent step. In the inspection method for inspecting the presence or absence of a linear stripe formed on the inspection object, image data for each pixel obtained by converting the imaging signal is input, and m rows n A change amount calculating step of emphasizing a vertical and horizontal density change amount by a first emphasis filter in a column; and a density change amount emphasized by the change amount calculating step is cut out by a grid including a plurality of data. An average calculation step of setting the average value of the plurality of data as an average change amount, and cutting out the change amount averaged in the average calculation step into a grid including a plurality of vertical and horizontal matrices, and adding the change amounts in the grid. Find the added value for each grid A change amount adding step, and p rows q with respect to the change amount obtained in the change amount adding step.
A defect determining step of detecting a streak by comparing a change amount emphasized in the vertical and horizontal density change amounts with the second threshold filter in the column and a predetermined threshold value, and comparing the change amount emphasized in the change amount calculating step with a predetermined threshold value. And a step of serially processing the change amount calculating step, the average calculating step, the change amount adding step, the change amount calculating step, and the defect determining step in this order. 2. The method according to claim 1, wherein the step of adding a change amount includes cutting out the plurality of data in a lattice shape such that the number of data in the transport direction of the inspection object is larger than the number of data in a direction orthogonal to the transport direction. 2. A change amount adding step of adding an amount of change of (i) to obtain an added value for each lattice.
The streak inspection method described. 3. A camera for picking up an image of the object to be inspected while transporting the sheet-like object to be inspected, and image data for each pixel obtained by converting an image pickup signal accompanying the image pickup are sent to a subsequent processing unit. And an image input unit that inputs image data sent from the image input unit, and emphasizes the amount of density change in the vertical and horizontal directions with respect to the image data by a first enhancement filter of m rows and n columns. The change amount calculation means and the average calculation means which cuts out the density change amount emphasized by the first change amount calculation means with a grid composed of a plurality of data and sets an average value of the plurality of cut out data as an average change amount. A microfilter unit, a change amount processed by the microfilter unit is cut out into a grid composed of a plurality of vertical and horizontal matrices, and a mura sector adder that obtains an addition value for each grid by adding the change amounts in the grid; For the sum obtained by the data adder, p rows q
A second emphasis filter in the column, a variation filter section comprising a variation calculation means for enhancing the density variation in the vertical and horizontal directions, and comparing the variation emphasized by the uneven filter with a predetermined threshold. A defect determination unit that detects a stripe and determines a defect, wherein the image input unit, the microfilter unit, the unevenness filter unit, and the defect determination unit are configured in series in this order. apparatus. 4. The mura sector adder of the mura filter section cuts out the plurality of data in a lattice shape such that the number of data in the transport direction of the object to be inspected is larger than the number of data in the direction orthogonal to the transport direction. 4. The stripe inspection apparatus according to claim 3, further comprising a non-uniform sector adder for obtaining an added value for each grid by adding the amount of change in the grid.