JPH07333173A - Defective pattern detection method and device therefor - Google Patents

Abstract

PURPOSE:To make a correction for an intermediate picture element value appearing due to the characteristics of a photographing system, as well as a correction for the superposing position of images, depending on the fine expansion, reduction, dislocation or the like of a pattern boundary CONSTITUTION:A hysteresis processing circuit 12 is provided for changing the value of a target picture element, using a value nearer the value of the element among maximum and minimum values within a window around the target picture element. Furthermore, a direction determination circuit 14 is provided to find a variation in X and Y directions for every picture element regarding an image (standard image) from a photographed image and a delay circuit 13, and identifying the dislocation of a photographed position from four types of patterns, together with a window selector circuit 18 for cutting out three types of 3X3 picture element windows from the photographed image, depending on the identified pattern. In addition, comparison circuits 15, 16 and 17 are provided to make a comparison between the cutout windows and images from the circuit 13, and evaluating a resulting difference, thereby selecting the minimum value of evaluation data from each of the circuits 15 to 17.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a defect pattern detecting method for detecting defects such as print omissions, foreign matter adhesion, and damages on an imaged object, for example, a pattern such as a pattern or characters or symbols printed on a print medium. Regarding the device.

[0002]

2. Description of the Related Art In a conventional defect pattern detecting apparatus, a plurality of identical print patterns are read at a constant cycle,
Overlay the read print pattern input image with the print pattern (standard pattern) image read one cycle before,
The method of extracting the different part was taken.

In this method of extracting the difference portion, since the latest image is always used as the standard pattern, it is easy to deal with the fluctuation of the illumination condition, the image shift, etc.
Since the relative position with respect to the immediately preceding image is only a problem, there is no problem that a large deviation occurs.

[0004]

However, in the boundary portion of the image, the brightness change has a sharp change in accordance with the change of the pattern, and therefore, when this is actually imaged, the pixel value is 2 It will take the mean value of the brightness of the two areas.

This intermediate value does not become a constant value between the standard image and the target image due to the optical blur of the image pickup system and the slight shift of the image forming position of the pattern pattern on the image pickup element, which causes an error at the time of comparison. Therefore, the conventional masking process or the like has been made so as not to compare the boundary portions, but this has a problem that it is not possible to accurately detect the defect at the image boundary portion. Also, the position of the pattern during printing is slightly different depending on the amount and viscosity of the ink,
This also makes it difficult to detect defects in fine patterns and pattern boundaries.

Further, in the conventional defect pattern detecting apparatus, a sheet on which a plurality of print patterns for inspecting a defect pattern are arranged is wound in a roll shape and set in the apparatus. Therefore, the sheet pulled out from the roll. When the print pattern arranged on the sheet is imaged in this warped state, a deviation occurs in the image pickup position from the standard pattern to be compared. In reality, the image pickup position is not a defective pattern. There was a problem that it was detected as a defective pattern.

Therefore, the present invention corrects an intermediate pixel value of each pixel, which occurs due to the characteristics of the image pickup system, which occurs at a fine pattern of a print pattern or a boundary portion of the pattern, and makes a delicate enlargement of the boundary portion of the pattern. It is an object of the present invention to provide a defect pattern detection method and apparatus capable of correcting the overlapping position of images in accordance with reduction, deviation, and the like.

[0008]

The invention according to claim 1 is
Detecting an image of an imaged object from light from the imaged object, comparing it with a standard image stored in advance, and detecting a difference between these images to detect a defective portion of the imaged image. In the imaging step of obtaining an image of the imaging target by the light from the imaging object, and for each pixel of the captured image obtained by this imaging step, the maximum value of the values of the peripheral pixels within the designated pixel number distance from the corresponding pixel and The distribution analysis step for obtaining the minimum value, and for each pixel, the value of the corresponding pixel is determined by the value closer to the value of the corresponding pixel obtained in the imaging step out of the maximum value and the minimum value obtained in this distribution analysis step. And a pixel value changing step of changing the pixel value.

According to a second aspect of the present invention, there is provided an image pickup means for detecting light from an image pickup object and a processing means for processing a signal from the image pickup means. The processing means is a pickup image obtained from the image pickup means. In a defect pattern detection device for detecting a defect pattern of a captured image by comparing a standard image set in advance and detecting a difference between these images, each pixel detected by the image capturing means is applicable. Distribution analysis means for obtaining the maximum and minimum values of the peripheral pixels within a specified number of pixel distances from the pixel, and for each pixel, detection by the imaging means of the maximum and minimum values obtained in this distribution analysis step And a pixel value changing means for changing the value of the corresponding pixel with a value closer to the value of the selected corresponding pixel.

According to a third aspect of the present invention, an image of the imaged object is detected from the light from the imaged object, compared with a standard image stored in advance, and a difference between these images is detected to detect the imaged image. In the method of detecting a defect pattern for determining a defective portion, the imaging step of obtaining an image of the imaging target by light from the imaging target, and the value of the corresponding pixel and the corresponding value for each pixel of the captured image obtained by this imaging step. An image-capturing change amount analysis step of obtaining a change amount between a pixel and a value of a neighboring pixel and its direction, and a change amount between a value of a corresponding pixel and a value of a neighboring pixel of the corresponding pixel for each pixel of the standard image. And a standard change amount analysis step of obtaining the direction thereof, a change amount and its direction obtained in the imaging change amount analysis step, and a change amount and its direction obtained in the standard change amount analysis step are compared to obtain a captured image. Standard image Those having a difference detection step of detecting a difference.

The invention according to claim 4 is provided with an image pickup means for detecting light from an image pickup object and a processing means for processing a signal from the image pickup means, and the processing means is a picked-up image obtained from the image pickup means. In a defect pattern detection device for detecting a defect pattern of a captured image by comparing a standard image set in advance and detecting a difference between these images, each pixel detected by the image capturing means is applicable. An image pickup change amount analyzing means for obtaining a change amount and a direction between a pixel value and a value of a neighboring pixel of the corresponding pixel, and a value of the corresponding pixel and a value of a neighboring pixel of the corresponding pixel for each pixel of the standard image Between the change amount and its direction, the change amount and its direction obtained by the imaging change amount analyzing means, and the change amount and its direction obtained by the standard change amount analyzing means. Comparison Te is obtained by providing a difference detecting means for detecting a difference between the captured image and the standard image.

The invention according to claim 5 is provided with an image pickup means for detecting light from an image pickup object and a processing means for processing a signal from the image pickup means, and the processing means is a pickup image obtained from the image pickup means. In a defect pattern detection device that detects a defect pattern of a captured image by comparing a preset standard image and detecting a difference between these images, the relative position of the captured pixel and the standard image is shifted. For a plurality of cases, the difference amount analysis means for obtaining the difference amount between the captured image and the standard image, and the difference amount evaluated by the difference amount analysis means that the difference amount is evaluated to be the smallest, A minimum difference amount selecting means for selecting is provided, and the defective portion of the captured image is detected based on the minimum difference amount selected by the minimum difference amount selecting means.

[0013]

In the present invention having such a structure, the image of the image pickup target is obtained by the light of the image pickup target. For each pixel forming this image, the maximum value and the minimum value of the values of the peripheral pixels within the specified number of pixel distances from the corresponding pixel are obtained, and the value of the corresponding pixel among the obtained maximum value and minimum value is determined. The value of the corresponding pixel is changed by the closer value.

Therefore, the intermediate value is eliminated, and the boundary portion of the print pattern becomes clear. Further, for each pixel, the change amount of the value of the corresponding pixel and the value of the neighboring pixel of the corresponding pixel and the direction thereof are obtained. Further, for each pixel of the standard image, the amount of change in the value of the pixel adjacent to the pixel of interest and the direction thereof are obtained.

The amount of change and the direction of each pixel obtained in the captured image with respect to the neighboring pixels are compared with the amount of change and the direction of each pixel obtained in the standard image with respect to the neighboring pixels. Differences are detected.

Therefore, this difference (analyzing this difference
), The shift of the image pickup position of the print pattern can be recognized, and the defect pattern of the print pattern is detected in consideration of the shift of the image pickup position.

Further, in a plurality of cases in which the relative positions of the captured image and the standard image are shifted, the difference amount between the captured image and the standard image is obtained, and the difference amount is the smallest among the obtained difference amounts. Those evaluated as are selected, and the defective portion of the captured image is detected based on the selected difference amount.

That is, by shifting the relative positions of the picked-up image and the standard image to cancel the shift of the picked-up position of the print pattern, the shift of the picked-up position is corrected and the defective pattern of the print pattern is detected.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing a schematic configuration of a defect pattern detection device to which the present invention is applied.

The printed material 2 wound in a roll is set on the supply side roller 1. In this printed matter, a plurality of identical print patterns to be inspected are arranged and printed at regular intervals. Alternatively, the printed matter 2 is configured by arranging a plurality of identical print patterns on a sheet at regular intervals.

A take-up side roller 2 is provided at a position separated from the supply-side roller 1 by a predetermined distance, and the printed matter 2 set on the supply-side roller is stretched around the take-up side roller 3 and the winding is performed. It is adapted to be wound by the take-up roller 3.

The supply side roller 1 and the winding side roller 3
A line sensor 4 as an image pickup means is installed on the printed material 2 stretched between the printed material 2 and an upper part of the supply side roller 3 so that the reflected light from the printed material 2 is incident on the line sensor 4. The illumination unit 5 is installed in the. The illumination driving unit 6 controls the supply of electric power to the illumination unit 5 to irradiate the surface of the printed material 2 with light. The line sensor includes a plurality of photoelectric conversion elements (for example, CCD (charge coupl
ed device) elements are arranged in one line (in a line).

The processing unit 7 processes the signal (imaging signal) output from the line sensor 4 to extract the defective portion of each print pattern of the printed matter, and the display unit 8 connected to this processing unit 7 The defective portion is displayed based on the processing result.

FIG. 2 is a block diagram showing a circuit configuration of essential parts of the processing section 7. The image pickup signal output from the line sensor 4 is an A / D (analog / digital) converter 11
Is input to the line sensor 4, converted into digital data for each pixel of the line sensor 4, and output to the hysteresis processing circuit 12.

As will be described later, the hysteresis processing circuit 12 performs optimum level correction on the input digital data. This hysteresis processing circuit 12
The output data (image data) is input to the delay circuit 13 and the direction determining circuit 14. The image data from the hysteresis processing circuit 12 delayed by one cycle from the delay circuit 13 is used as the direction determining circuit 14, the first comparison circuit 15, the second comparison circuit 16, and the third comparison circuit 1.
It is designed to be input to 7.

As will be described later, the direction determining circuit 14 specifies a comparison direction for the image data directly input from the hysteresis processing circuit 12, and the image data input from the delay circuit 13 for each comparison direction. Overlay is performed and the result of the overlay is compared to determine the direction type.

As will be described later, the window switching circuit 18 cuts out three types of windows according to the direction type determined by the direction determining circuit 14. For each of the three types of windows cut out by the window switching circuit 18, one cycle before input from the delay circuit 13 by the first comparison circuit 15, the second comparison circuit 16, and the third comparison circuit 17, respectively. The amount of difference is obtained by superimposing it on the image data, and the amount of difference is evaluated and converted into evaluation value data. That is, the evaluation value data is data indicating a defective portion of the image data.

The evaluation value data obtained by the comparison circuits 15, 16 and 17 are output to the minimum value selection circuit 19 as the minimum difference amount selection means. This minimum value selection circuit compares each evaluation value data, selects the data having the smallest evaluation value, and outputs it to the binarization circuit 20.

The binarization circuit 20 binarizes the evaluation value data with a predetermined threshold value and outputs it to the noise removal circuit 21. The noise removing circuit 21 is a circuit for reducing / enlarging the binarized data to remove the noise portion of about one pixel.

The binarized data from which the noise portion is removed by the noise removing circuit 21 is the data obtained by extracting the final defective portion, and the defective portion is displayed on the display 8 based on this data.

FIG. 3 is a block diagram showing a circuit configuration of a main part of the hysteresis processing circuit 12. The digital data output from the A / D converter 11 is input to the window cutout circuit 31.

In the window cutout circuit 31, as will be described later, for each digital data (pixel data, respectively referred to as target pixel data), 3 × 3 pixels (nine neighboring data including the target pixel) Alternatively, window cutout of 5 × 5 pixels (25 neighboring data including the target pixel) is performed. It should be noted that although the description has been given here because a practically sufficient effect can be obtained with 3 × 3 pixels or 5 × 5 pixels, the present invention is not limited to this, and other window cutouts of, for example, 7 × 7 pixels are provided. There is also a method. Hereinafter, for simplification, an example of cutting out a window of 3 × 3 pixels will be described.

The data of 3 × 3 pixels cut out for each digital data is output to the maximum value extraction circuit 32 and the minimum value extraction circuit 33, and the digital data of the initial value is input to the maximum / minimum selection circuit 34. It is output as data.

The maximum value extraction circuit 32 extracts the maximum value data from the 3 × 3 pixel data (nine pieces each) cut out for each digital data, and performs the maximum / minimum selection. Output to the circuit 34. The minimum extraction circuit 33 also extracts the minimum value data from the 3 × 3 pixel data cut out for each digital data and outputs it to the maximum / minimum selection circuit 34.

The maximum / minimum selection circuit 34 selects the data having the smaller difference between the maximum value data and the minimum value data from the initial value data, and selects the selected data as digital data of the corresponding pixel. Output as.

The maximum value extraction circuit 32 and the minimum value extraction circuit 33 constitute distribution analysis means, and the maximum / minimum selection circuit 34 constitutes pixel value changing means. Figure 4
FIG. 3 is a block diagram showing a circuit configuration of a main part of the window cutout circuit 31.

The digital data output from the A / D converter 11 is input to the first shift register 41 in which registers for storing digital data are serially connected by the number of digital data output from the line sensor 4. And a first 3-connected shift register 4 in which three registers for sequentially storing digital data are serially connected.
Input to 2.

Further, the first shift register 41 is serially connected with a second shift register 43 in which registers for storing digital data are serially connected by the number of digital data output from the line sensor 4. At the same time, the second 3-connected shift register 44, in which three registers for storing digital data are serially connected, is also serially connected.

Further, the second shift register 43 is serially connected with a third 3-connected shift register 45 in which three registers for storing digital data are serially connected.

Each of the three-connected shift registers 42, 44,
The stored digital data is output from each of the registers 45. The digital data from the register in which digital data is first stored from the A / D converter 11 of the first 3-connected shift register 42 is G1.
And digital data from the registers that are sequentially shifted and stored are G2 and G3. Also, the digital data from the register in which the digital data is first stored from the first shift register 41 of the second 3-connected shift register 44 is set to G4, and the digital data from the register sequentially shifted and stored is G5. , G6. Further, the second of the third 3-connected shift register 45 is
The digital data from the register in which the digital data is first stored from the shift register 43 is designated as G7, and the digital data from the registers which are sequentially shifted and stored are designated as G8 and G9.

Then, as shown in FIG. 5, a 3 × 3 pixel window (image data) of the image data is cut out. Therefore, the digital data of G1 to G9 is output to the maximum value extraction circuit 32 and the minimum value extraction circuit 33,
Digital data of G5 which is digital data (pixel data of interest) at the center of a 3 × 3 pixel window of image data is output to the maximum / minimum selection circuit 34 as initial value data.

FIG. 6 is a block diagram showing a circuit configuration of a main part of the maximum / minimum selection circuit 34. The initial value data (G5) output from the window cutout circuit 31 and the maximum value data output from the maximum value extraction circuit 32 are input to a maximum value difference calculation circuit 51, and this maximum value difference calculation circuit The data difference is calculated at 51.

The initial value data from the window cutout circuit 31 and the minimum value data output from the minimum value extraction circuit 33 are input to the minimum value difference calculation circuit 52, and the minimum value calculation circuit 52 outputs the data. The data difference is calculated.

The data differences calculated from the maximum value difference calculation circuit 51 and the minimum value difference calculation circuit 52 are input to the comparison circuit 53, and the comparison circuit 53 compares the two data differences and outputs the smaller data difference. One is designated to the selection circuit 54. The maximum value data and the minimum value data are input to the selection circuit 54 from the maximum value extraction circuit 32 and the minimum value extraction circuit 33, respectively. One of the value data and the minimum value data is selected and output as digital data of the corresponding pixel.

That is, for simplification, a one-dimensional example will be described. Even though the same print pattern is imaged, as shown in FIG. 7A and FIG. 7B, the print pattern When the image pickup position shifts, there is a problem that the value of intermediate digital data (pixel data) shifts. The image pickup position is (i-
If it is shifted in the direction of the 1) th pixel, digital data can be obtained as shown in FIG. 7B. Here, the digital data values of the i-th pixel and the (i + 1) -th pixel change by d1 and d2. In order to solve this problem, the hysteresis circuit 12 configured as described above replaces the intermediate digital data with the value of the digital data in the vicinity, and eliminates the value.

As shown in FIG. 8 (a), if digital data is obtained, first, a window is cut out from the three pixel data before and after, including the pixel of interest. Regarding the window of the i-th pixel, the digital data value L1 of the (i-1) -th pixel, the digital data value (initial value) L2 of the i-th pixel, and the digital data value of the (i + 1) -th pixel If L3, the maximum value of this window is L
3, the minimum value L1, and the initial value L2. So L3-L
2, L2-L1 is calculated and L3-L2> L2-L1 is established, so the value of the digital data of the i-th pixel, which is the target pixel, is changed to L1.

Further, for the window of the (i + 1) th pixel, the digital data value L2 of the ith pixel, (
i + 1) Digital pixel value of pixel (initial value) L
Assuming that the value of the digital data of the 3rd (i + 2) th pixel is L4, the maximum value of this window is L4, the minimum value is L2,
The initial value is L3. Therefore, L4-L3 and L3-L2 are calculated, and L4-L3 <L3-L2, so the value of the digital data of the (i + 1) th pixel, which is the pixel of interest, is changed to L4.

As a result, as shown in FIG. 8B, the image data without intermediate digital data can be changed. That is, the boundary portion such as the pattern in the print pattern becomes clear.
Although a one-dimensional example has been described here for simplification, the same processing is performed in a two-dimensional manner as shown in FIG. 5, for example, so that a boundary portion such as a pattern in a print pattern becomes clear.

FIG. 9 is a block diagram showing a circuit configuration of a main part of the direction determining circuit 14. The output data (image data) from the hysteresis processing circuit 12 includes an X-direction change amount dxr extraction circuit 61 and a Y-direction change amount dyr extraction circuit 62.
Entered in. Further, the output data (image data) delayed by one cycle from the delay circuit 13 is the change amount d in the X direction.
It is input to the xs extraction circuit 63 and the Y-direction change amount dys extraction circuit 64.

Circuits corresponding to the window cutout circuit 31 of the hysteresis processing circuit 12 are incorporated in the extraction circuits 61, 62, 63 and 64, and the following equations (1) and (2) are respectively provided. The amounts of change dxr, dxs and dyr, dys are calculated by the equations.

Dxr (dxs) = (G3 + G6 + G9) − (G1 + G4 + G7) ... (1) dyr (dys) = (G1 + G2 + G3) − (G7 + G8 + G9) ... (2) The X direction addition circuit 65 is the x direction change circuit 65, and the x direction addition circuit 65 is the x direction change circuit 65. 61 and the change amounts dxr and dxs output from the X-direction change amount dxs extraction circuit 63 are added, and the addition result dx = d
It outputs xr + dxs to the direction determination circuit 67.

The Y-direction addition circuit 66 adds the dy and dys output from the Y-direction change amount dyr extraction circuit 62 and the Y-direction change amount dys extraction circuit 64, and outputs the addition result dy = dyr + dys. Output to the direction determination circuit 67.

The X-direction change amount dxr extraction circuit 61 and the Y-direction change amount dyr extraction circuit 62 constitute an imaging change amount analysis means, and the X-direction change amount dxs extraction circuit 63 and the Y-direction change amount d.
The ys extraction circuit 64 constitutes a standard change amount analysis means.

The direction judging circuit 67 outputs d output from the X-direction adding circuit 65 and the Y-direction adding circuit 66.
The types of image shift based on x and dy are shown below.
Determine from the types of types.

(Type 0) is when the absolute values of dx and dy are within α times that of the other party, and the sign of dx × dy is negative. (Type 1) is when the absolute value of dy is α times or more the absolute value of dx.

(Type 2) is when the absolute values of dx and dy are within α times that of the other party, and the sign of dx × dy is positive. (Type 3) When the absolute value of dx is α times or more the absolute value of dy.

Here, α has an optimum value of about 2 to 3. These four types 0 to 3 correspond to image shifts as shown in FIG. That is, FIG. 10 (a), FIG.
(b), FIG. 10 (c), and FIG. 10 (d), assuming that the vertical direction is the y-axis and the horizontal direction is the x-axis, the type 0 is shown in FIG. 10 (a).
, Type 1 corresponds to FIG. 10 (b), and type 2
Corresponds to FIG. 10 (c), and type 3 corresponds to FIG. 10 (d).

That is, the direction determining circuit 14 is shown in FIG.
As shown in (a), if the print pattern (the boundary of the print pattern, indicated by the solid line) is misaligned with the image data captured one cycle before (shown by the broken line), the pattern brightness of the print pattern Although the position is one pixel or not shown (for example, it can be shifted by 2 pixels by cutting out a window of 7 × 7 pixels) in the direction of change in height, overlapping is performed while shifting several pixels and the one with the smallest difference is detected. Is done to As a result, in the image data just captured, the captured print pattern moves from the broken line to the solid line, as shown in FIG. 11 (b). Δf and θ in FIG. 11 (b) are calculated by the equations (3) and (4).

[0059]

[Equation 1]

The window switching circuit 18 cuts out a window of 5 × 5 pixels, and the direction determining circuit 1
According to the type determined by 4, three types of 3 × 3 pixel windows are cut out at respective positions of A, B, and C in the 5 × 5 pixel window.

The three types of cut-out 3 × 3 pixel windows A, B, and C are the first comparison circuit 1 respectively.
5, supplied to the second comparison circuit 16 and the third comparison circuit 17. On the other hand, from the delay circuit 13 to each comparison circuit 15,
Image data of one cycle before is supplied to 16 and 17, and in each of the comparison circuits 15, 16 and 17, the windows A, B and C and the image data of one cycle before are overlapped with each other, and the difference amount thereof is calculated. Is evaluated.

Each of the comparison circuits 15, 16 and 17 outputs the evaluation value data to the minimum value selection circuit 19, respectively. The method of evaluating the difference amount is calculated as described below.

Each window of the 3 × 3 pixel window of the image data of one cycle before output from the delay circuit 13 (
) Assume that the data is G1 to G9, and each of the 3 × 3 pixel windows (of brightness) output from the window switching circuit 18
When the data are g1 to g9, the evaluation value P0 as a simple brightness evaluation is calculated by the equation (5).

[0064]

[Equation 2] Further, the evaluation value P1 when the difference amount of the print pattern including the direction is strictly evaluated is calculated by the equation (6) using dxs, dys, dxr, and dyr.

[0065]

[Equation 3]

The subscript i of dxs, dys, dxr, and dyr
Indicates the position of a 3 × 3 pixel window.
When the judgment formula D calculated by the formula (7) is close to 0, the evaluation value P0 according to the formula (5) is used, and in other cases, the evaluation value P1 according to the formula (6) is used. It can also take a mixed form.

Each of the comparison circuits 15, 16 and 17 has (6
When the evaluation value data is obtained by the equation), each of the comparison circuits 15, 16 and 17 constitutes a difference detecting means. In the present embodiment having such a configuration, the image data captured by the line sensor 4 is input to the hysteresis processing circuit 12 via the A / D converter 11, and in the hysteresis processing circuit 12, 3 × 3 pixels are displayed. The window of is cut out, the maximum value and the minimum value of the nine pixels around the target pixel including the target pixel are obtained, and the one of the maximum value and the minimum value that is closer to the value of the target pixel is selected, The selected value is set as the value of the pixel of interest.

As a result, the intermediate data at the boundary of the print pattern is removed, and the boundary is clearly corrected. The image data whose boundary is clearly corrected in this way is compared with the pixel data imaged one cycle before by the direction determination circuit 14, and the direction of the image shift is determined from among the four types shown in FIG. It is determined.

According to the type determined by the direction determining circuit 14, the window switching circuit 18 extracts three types of 3 × 3 pixel windows A, B and C,
The windows A, B, and C are compared with the image data of one cycle before output from the delay circuit 13 by the first comparison circuit 15, the second comparison circuit 16, and the third comparison circuit 17, respectively, and the difference amount thereof. Is evaluated.

The minimum value selection circuit 19 includes the comparison circuits 15,
Of the three types of evaluation data output from 16 and 17, the data having the smallest evaluation value is selected to select the binarization circuit 20.
Output to.

The evaluation data binarized by the binarization circuit 20 is output to the noise removal circuit 21, and the noise removal circuit 21 removes noise. A defect pattern is extracted from the evaluation data from which noise has been removed by the noise removal circuit 21.

As described above, according to the present embodiment, the maximum value and the minimum value in the window constituted by the pixel of interest and the peripheral pixels are obtained, and the maximum value and the minimum value are close to the value of the pixel of interest. The hysteresis processing circuit 12 that changes the value of the pixel of interest with the other value, the X direction change amount and the Y direction change amount for each pixel of the captured image, and 1 from the delay circuit 13.
The amount of change in the X direction and the amount of change in the Y direction are obtained for each pixel of the image before the cycle (standard image), and the shift in the image pickup position of the print pattern is determined from the four types of patterns based on these amounts of change. The direction determining circuit 14, the window switching circuit 18 that cuts out three types of 3 × 3 pixel windows from the captured image according to the determined pattern, and the delay circuit 13 for each of the three types of 3 × 3 pixel windows.
Comparing circuits 15 and 16 and 17 for evaluating the difference amount, and a minimum value selecting circuit 19 for selecting the minimum value of the evaluation data output from the comparing circuits 15, 16 and 17. By providing the above, the hysteresis processing circuit 12 can correct the intermediate data, and the direction determining circuit 14 changes the amount of change in the X direction and the Y direction for each pixel of the captured image and the image from the delay circuit 13. X after each pixel
The positional deviation pattern can be determined by comparing the amounts of change in the direction and the Y direction.

Therefore, the hysteresis processing circuit 12 can eliminate the unclear pixel value and make the image pattern clear. Further, the direction determining circuit 14 can determine the direction of the positional deviation of the print pattern.
As a result, it has been necessary to consider shifts in eight directions in the conventional print pattern, whereas in this embodiment, it is sufficient to consider only one direction. Ie 1
Since only the comparison circuit in the direction is required, the circuit configuration can be simplified and the positional deviation at the time of capturing the print pattern can be accurately corrected.

As described above, according to this embodiment, the defect pattern can be detected with high accuracy. In addition, in this embodiment, the reflected light from the print pattern is detected, and the defect pattern is detected. However, the present invention is not limited to this. For example, transmitted light from the print pattern is detected. It can also be applied to what is done.

[0075]

As described above in detail, according to the present invention,
Occurs at the fine pattern of the print pattern or at the boundary of the pattern,
It is possible to correct the intermediate pixel value of each pixel that occurs due to the characteristics of the imaging system, and to correct the overlapping position of images according to subtle enlargement, reduction, misalignment of the pattern boundary. A detection method and an apparatus therefor can be provided.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of a defect pattern detection device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a circuit configuration of a main part of a processing unit of the defect pattern detection apparatus of the embodiment.

FIG. 3 is a block diagram showing a circuit configuration of a main part of a hysteresis processing circuit of the defect pattern detection apparatus according to the embodiment.

FIG. 4 is a block diagram showing a circuit configuration of a main part of a window cutout circuit of the defect pattern detection apparatus of the embodiment.

FIG. 5 is a view showing a window of 3 × 3 pixels cut out by a window cutting circuit of the defect pattern detection apparatus of the embodiment.

FIG. 6 is a block diagram showing a circuit configuration of essential parts of a maximum / minimum selection circuit of the defect pattern detection apparatus of the embodiment.

FIG. 7 is a diagram showing a one-dimensional distribution of imaged pixel data of the embodiment.

FIG. 8 is a diagram showing a one-dimensional distribution of pixel data before and after processing by the hysteresis processing circuit of the defect pattern detection apparatus of the embodiment.

FIG. 9 is a block diagram showing a circuit configuration of a main part of a direction determining circuit of the defect pattern detecting apparatus of the embodiment.

FIG. 10 is a diagram showing four types of patterns determined by a direction determining circuit of the defect pattern detecting apparatus of the same embodiment.

FIG. 11 is a diagram showing a direction determining circuit of the defect pattern detecting apparatus of the embodiment and a diagram showing before and after image shift processing by the circuit.

[Explanation of symbols]

 4 ... Line sensor, 7 ... Processing part, 12 ... Hysteresis processing circuit, 13 ... Delay circuit, 14 ... Direction determination circuit, 15, 16, 17 ... Comparison circuit, 18 ... Window switching circuit, 19 ... Minimum value selection circuit, 31 … Window cutting circuit.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G06T 5/00 // B41J 29/46 C 0834-5H G06F 15/70 330 N

Claims (5)

[Claims] 1. A defective portion of the captured image is detected by detecting an image of the captured object from light from the captured object, comparing it with a standard image stored in advance, and detecting a difference between these images. In the method of detecting a defect pattern to be determined, an image capturing step of obtaining an image of the image capturing target by light from the image capturing target, and each pixel of the captured image obtained by this image capturing step within a designated pixel number distance from the corresponding pixel. A distribution analysis step of obtaining the maximum value and the minimum value of the values of a certain peripheral pixel,
For each pixel, a pixel value changing step of changing the value of the corresponding pixel with a value closer to the value of the corresponding pixel obtained in the imaging step out of the maximum value and the minimum value obtained in the distribution analysis step. A method for detecting a defect pattern, comprising: 2. An image pickup means for detecting light from an image pickup target, and a processing means for processing a signal from the image pickup means,
The processing means compares a captured image obtained from the image capturing means with a preset standard image and detects a difference between these images to detect a defect pattern of the captured image. In the apparatus, for each pixel detected by the image pickup means, a distribution analysis means for obtaining the maximum value and the minimum value of the values of peripheral pixels within a designated pixel number distance from the corresponding pixel, and this distribution analysis for each pixel And a pixel value changing means for changing the value of the corresponding pixel with a value closer to the value of the corresponding pixel detected by the image pickup means out of the maximum value and the minimum value obtained in the step. Device for detecting defective patterns. 3. A defective portion of the picked-up image is determined by detecting an image of the picked-up object from light from the picked-up object, comparing it with a standard image stored in advance, and detecting a difference between these images. In the method of detecting a defect pattern, the image pickup step of obtaining an image of the image pickup object by the light from the image pickup object, and for each pixel of the image pickup image obtained by this image pickup step, the value of the corresponding pixel and the vicinity of the corresponding pixel An image pickup change amount analysis step of obtaining a change amount between the pixel value and the direction thereof, and a change amount between the value of the corresponding pixel and a value of a neighboring pixel of the corresponding pixel for each pixel of the standard image and its The standard change amount analysis step of obtaining a direction, the change amount and its direction obtained in the imaging change amount analysis step are compared with the change amount and its direction obtained in the standard change amount analysis step, and the captured image is obtained. And before And a difference detecting step of detecting a difference from the standard image. 4. An image pickup unit for detecting light from an image pickup target and a processing unit for processing a signal from the image pickup unit,
The processing means compares a captured image obtained from the image capturing means with a preset standard image and detects a difference between these images to detect a defect pattern of the captured image. In the device, for each pixel detected by the image pickup means, an image pickup change amount analysis means for obtaining a change amount and a direction between a value of a corresponding pixel and a value of a neighboring pixel of the corresponding pixel, and the standard image For each pixel, standard change amount analysis means for analyzing the change amount and the direction between the value of the corresponding pixel and the value of the neighboring pixel of the corresponding pixel, and the change amount obtained by the imaging change amount analyzing means and its A defect pattern characterized by providing a difference detecting means for comparing a direction with a change amount obtained by the standard change amount analyzing means and its direction to detect a difference between the captured image and the standard image. Detection device. 5. An image pickup unit for detecting light from an image pickup target, and a processing unit for processing a signal from the image pickup unit,
The processing means compares a captured image obtained from the image capturing means with a preset standard image and detects a difference between these images to detect a defect pattern of the captured image. In the apparatus, for a plurality of cases in which the relative positions of the image pickup pixel and the standard image are shifted, a difference amount analysis unit that obtains the difference amount between the captured image and the standard image, and the difference amount analysis unit are obtained. And a minimum difference amount selecting means for selecting a difference amount evaluated to be the smallest among the plurality of difference amounts, and based on the minimum difference amount selected by the minimum difference amount selecting means, the imaging is performed. A defect pattern detection device characterized by detecting a defective portion of an image.