JP2019215659A - Image inspection device and image detection method thereof - Google Patents

Abstract

To perform a combination, in a block image divided vertically and horizontally, so that the number of alignment points for correction of a positional deviation satisfies the necessary number, thereby correcting the positional deviation of an inspection object.SOLUTION: An image inspection device includes a block image correction unit, detects a POS point as an alignment point from a reference image including an inspection object, and defines a point at which a positional deviation amount can be calculated for pixels in the vicinity of an area corresponding to the POS point for each block image obtained by dividing an inspection image including the same inspection object vertically and horizontally as a POM point. If the number of POM points in one block image is less than a set number, the adjacent block images are combined until the number of POM points exceeds the set number. An average value is obtained by summing the positional deviation amounts of the uncombined or combined block images and dividing the sum by the number of POM points, and the sign is inverted to obtain a registration correction amount. An image block of the comparison image including the inspection image corrected by the alignment correction amount is compared with a block image of the reference image to determine whether the inspection image is good or bad.SELECTED DRAWING: Figure 1

Description

  An embodiment of the present invention relates to an image inspection apparatus for an inspection object and an image detection method thereof.

  Conventionally, the inspection of the quality of the characters to be printed is, for example, using a reference image including non-defective reference characters as a criterion, by comparing whether the read print characters to be inspected match the reference characters, Good / bad print characters are determined. Characters, symbols, and pictures printed on the work or on a label attached to the work are optically read by an imaging unit such as a line sensor, and are generated as still image data (video signal) by photoelectric conversion. .

  Since the line sensor has a configuration in which a plurality of image pickup elements are arranged in a line, the line sensor itself or a workpiece is moved to perform imaging by scanning. For example, when the work is a cylindrical resin tube, in order to read a character printed on a side surface of the resin tube or a label printed thereon, the resin tube is rotated and read. At this time, if there is an eccentricity of the rotation axis of the resin tube or a shift in the sticking direction, the read image may be tilted in the image data and may be shifted from the reference character. .

Japanese Patent No. 3725207

  In the case where the inclination or the like generated in the read inspection image data described above is determined by comparison with a reference, it is necessary to perform misregistration correction. For example, in Japanese Patent Application Laid-Open No. H11-163, the read image data (pixel group) is divided into a fixed number of line block images, and positional deviation correction is performed in block units. In order to perform position correction, a plurality of feature points must be included in each line block image, and it is difficult to set image segmentation depending on the image to be inspected.

  In addition, there is also displacement correction using band-shaped line block images separated at equal intervals in the horizontal direction of the image. However, as the length of the band-shaped line block image increases, the amount of displacement correction on the end side increases. The amount of arithmetic processing increases, and higher performance for realizing faster processing time is desired. In a strip-shaped line block image, for example, when an image is read by a line sensor, an inspection image read from a label having a middle bend or the like may be displaced vertically in the middle of a character string. It is not possible to cope with partial displacement correction such that a part of a character string described in a row is vertically displaced near the center of one strip-shaped line block image.

  Therefore, the present invention provides an image inspection apparatus that performs alignment so that alignment points for misalignment correction satisfies a set number in a vertically and horizontally divided block image including an inspection object, and performs misalignment correction of the inspection object, and An object of the present invention is to provide an image detection method.

  An image inspection apparatus according to an embodiment of the present invention provides a reference image including an inspection target with coordinate information, an alignment point detection unit that detects an alignment point from the reference image, and vertically and horizontally divides an inspection image including the inspection target. In the block image obtained, the density of the pixel at the alignment point is compared with the density of a pixel near the coordinates corresponding to the alignment point in the block image, and the amount of positional shift due to the difference is calculated. The amount calculation unit, and a pixel from which the amount of displacement can be calculated is a point at which the amount of displacement can be calculated. If the number of points at which the amount of displacement can be calculated is less than a set number in one block image, A block combining unit that repeats combining with an adjacent block image until the set number is exceeded, and a point at which the displacement amount exceeding the set number can be calculated with the inspection target of the reference image as a reference position. A correction amount detection unit that calculates an alignment correction amount that corrects a positional shift of the inspection target of the inspection image, and a block image of the inspection image corrected with the alignment correction amount, and the reference image. Are compared to determine whether the inspection image is good or bad.

  According to the present invention, there is provided an image inspection apparatus that performs alignment so that alignment points for positional deviation correction satisfies a set number in a vertically and horizontally divided block image including an inspection target and corrects positional deviation of an inspection target, and , The image detection method can be provided.

FIG. 1 is a diagram conceptually illustrating a configuration example of an image inspection apparatus according to an embodiment. FIG. 2 is a diagram for explaining the block image correction. FIG. 3 is a diagram for explaining image combination for including a registration point in a block image. FIG. 4 is a flowchart for explaining image position shift correction by the image inspection apparatus of the embodiment. FIG. 5 is a diagram for describing image inspection using a block image for a color image.

Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.
The image inspection apparatus according to the present embodiment includes a block image obtained by vertically and horizontally dividing a reference image including a reference character and a reference pattern to be inspected, and a block obtained by vertically and horizontally dividing a captured inspection image including the same inspection object in the same section as the reference image. Divide into images.

  The image inspection device includes a block image correction unit. The block image correction unit performs, for each block image, an alignment point (POS point), which is a feature point set on coordinates provided in advance in the reference image, and a vicinity of an area corresponding to the POS point on the inspection image. That is, by performing so-called template matching, the amount of displacement is calculated for each of the X direction and the Y direction of the coordinates. In the following description, the alignment point at which the amount of misalignment has been calculated is referred to as a POM point. If the number of POM points in one block image is less than the preset number, the image is combined with the adjacent block images until the number exceeds the preset number. For block images with a POM point equal to or greater than a predetermined number, including the combined block images, the position shift amount is summed, divided by the number of POM points to obtain the average value, and the sign is inverted to reverse the sign to correct the position alignment Amount. An image block of a comparative image composed of the inspection image corrected by the alignment correction amount is compared with a block image of the reference image to determine whether the inspection image is good or bad.

[First Embodiment]
FIG. 1 is a diagram conceptually illustrating a configuration example of the image inspection apparatus according to the first embodiment.
The image inspection apparatus 100 includes an imaging unit 1, an image processing unit 2, a control unit 3, a display unit 4, an input unit 5, and a storage unit 6.
In the present embodiment, the imaging unit 1 optically reads printed characters, symbols, and patterns, and generates still image data (digital signals) by photoelectric conversion.

  The imaging unit 1 captures a still image in a frame unit by a so-called digital camera using a CCD imaging device or a CMOS imaging device, or captures an image using a line sensor in which the imaging devices are arranged in a line. In the present embodiment, the imaging unit 1 using a line sensor is used. The line sensor captures images of characters, symbols, and pictures by moving the line sensor itself or the inspection target (including translation and rotation) and scanning. In particular, the imaging unit 1 using a line sensor is suitable for reading characters on a label printed or pasted on a curved surface.

The display unit 4 uses, for example, a liquid crystal display monitor. The display unit 4 displays various information such as an operation screen, an inspection status, and a determination result according to the control of the control unit 3. The input unit 5 uses, for example, various switches, input devices such as a keyboard and a mouse in combination. As the input unit 5, a touch panel may be provided on the screen of the display unit 4 and used.
The storage unit 6 stores a program for operating the control unit 3 and application software. Further, it also stores coordinate information of a positioning point, which is a feature point described later, a position correction amount of a block image, and the like.

  The image processing unit 2 performs CDS (correlated double sampling) processing, AGC (gain control) processing, and the like on image data (reference image REM, inspection image DI) output from the imaging unit 1. At this time, the density (lightness) of each pixel is digitized into a lightness value (gray scale) of 256 gradations by quantization. This lightness value is a pixel value. The image processing unit 2 outputs the image data of the reference image to the alignment point detection unit 16 and the storage unit 6, and outputs the image data of the inspection image to the block division unit 11. The image processing unit 2 may be configured integrally with the control unit 3.

  The control unit 3 includes a block division unit 11, a block image correction unit 12, a comparison image generation unit 13, a comparison unit 14, a determination unit 15, and an alignment point detection unit 16. The control unit 3 can be configured by a general-purpose personal computer by using a dedicated program or application software.

  The block division unit 11 gives the coordinate information (XY coordinates) to the image data (one frame) of the reference image and the inspection image output from the image processing unit 2, and outputs the reference image 31 (see FIG. 2). REM) and a frame of the inspection image 32 (DI) are divided vertically and horizontally to generate a plurality of block images.

  The position and size of the block image on the frame are defined by the coordinate information, and the coordinate information is also given to the inspection target in the block image. In the following description, each of the partial images divided vertically and horizontally may be referred to as a block image, or may be referred to as a sub-line block image because a line block divided into a band shape is divided vertically. After giving the same coordinate information to the reference image and the inspection image, the block division unit 11 performs vertical and horizontal division in the same section to generate a block image. In this division, it is assumed that the same inspection target is included in the reference image and the inspection image. In FIG. 2, the frame part 41 is an inspection target.

  The block image correction unit 12 determines, for each block image, an alignment point (POS point), which is a feature point set on coordinates provided in the reference image, and a POS point on the inspection image. By performing so-called template matching on the vicinity of the corresponding area, the amount of displacement is calculated for each of the X and Y directions of the coordinates. If the point (POM point) at which the amount of displacement can be calculated is less than the predetermined number, the block image is combined with an adjacent block image and the POM points of the combined block image are added up. Until the sum of the POM points exceeds the set number, the POM points are continuously combined with the adjacent block images. For a block image in which the number of POM points is greater than the set number (non-combined) even if not combined, the amount of displacement of the POM points in the block is summed for each of the X and Y coordinates and divided by the number of POM points to obtain the average value By reversing the obtained code, the block image is set as the position correction amount in the X direction and the Y direction, respectively. In the combined block image, the POM points included in all the combined blocks are used to calculate the alignment correction amount in the same manner as described above.

  The alignment point detection unit 16 sets coordinates (coordinate information) in the reference image. Next, a positioning point (POS point) formed by the minimum pixel group shown in FIG. 3 is detected from the reference image in order to perform positioning. The alignment points used for comparing the images are the same as the characteristic points, and may be a known pattern such as a characteristic pattern including a character or a ruled line, an edge thereof, or a luminance different from the surroundings. The registration point detection unit 16 creates map data by associating the registration points (POS points) with the position information (coordinate information) of each block image of the reference image, and stores the map data in the storage unit 6.

The block image correction unit 12 includes a position shift amount calculation unit 21, a block combination unit 22, and a correction amount calculation unit 23.
The displacement calculator 21 calculates the displacement around the POS point by template matching. As a calculation method, the density of the pixel at the POS point in the reference image is compared with the density of the pixel near the coordinates corresponding to the POS point in the inspection image. This comparison, that is, the total value of the absolute values of the differences in density for each pixel is calculated. Based on the magnitude of the total value, the displacement amount between the POS point and the pixel (point) near the coordinates corresponding to the POS point in the inspection image is calculated.

  Points at which the amount of displacement can be calculated are recorded as POM points (points at which the amount of displacement can be calculated) in the block image. In addition, when the image is significantly displaced, or when the symbol that should exist at the POS point position is missing, the difference between the compared images is large and the total value of the differences does not become sufficiently small. Not accepted as a POM point.

The block combining unit 22 integrates the number of POM points in one block image of the inspection image based on the calculation result of the displacement amount calculating unit 21. Also, the presence or absence of four or more POM points in one block image of the inspection image is determined, and the POM points are combined with adjacent block images until the total number of POM points becomes the set four or more.
In each block image cut out by division, four or more POM points are not always calculated. Therefore, if a block image has less than four POM points, it is combined with an adjacent block image until the total number of POM points becomes four or more. In the present embodiment, four POM points are set as an example. However, the number is not limited to four. Depending on the shape, image quality, resolution, and the like of the inspection target in the inspection image, The number is set as appropriate.

  This will be described with reference to FIG. Here, a set number of block images is four as an example. In the block image SB1, four block images P1, P2, P3, and P4 are arranged in series. In this example, each of the block images P1, P2, and P4 has one POM point that can be used for alignment. The block image P3 has four POM points that can be used for alignment. Therefore, the block images P1, P2, and P4 have insufficient number of POM points to perform the alignment correction.

  First, as shown in the block image SB2, the block image P1 and the block image P2 are combined to form a block image P12. Although this block image P12 has two POM points, it is still insufficient for the set number. On the other hand, the block image P3 and the block image P4 are combined to form a block image P34. There are five POM points in this block image P34, which satisfies four or more conditions.

  However, since the number of POM points is still insufficient for the set number, the block image P12 is combined with the adjacent block image P34 to form the block image P as shown in the block image SB3. In this block image P, there are seven POM points, which satisfies four or more conditions.

  In addition, in combining adjacent block images, a priority and a combining order may be set. For example, as a first engagement order, one set is combined and sequentially combined from one end of the arrangement of the block images to, for example, the left end from the right end block image. As a second combination order, when a block image is adjacent to both sides, the block image is preferentially combined with the block image having the larger number of POM points. As a third combination order, the block images having the largest (or smallest) POM point are sequentially combined so as to spread to both sides.

  The correction amount calculation unit 23 uses the POM point position shift amount information in the uncombined or combined block image in which the number of POM points of the inspection image satisfies the condition of four or more to inspect the reference image. The amount of correction in the X and Y directions is calculated so that the position of the same specific part of the inspection target in the inspection image approaches the position of the specific part. The correction amount is calculated by summing the displacement amounts of the POM points in the block for each of the X and Y coordinates, dividing by the number of POM points, obtaining an average value, and multiplying the obtained value by -1.

  As a specific example, the frame portion 41 to be inspected existing in the block image 42a of the reference image 31 (REM) shown in FIG. 2 is set as the reference position. An alignment correction amount for the position of the frame portion 41 of the block image 43a of the inspection image 32 (DI) to approach the position of the frame portion 41 of the reference image 31 is calculated.

  As shown in FIG. 2, the comparison image generation unit 13 performs correction (movement in the X and Y directions) on each of the block images 43 a to 43 n of the inspection image 32 using the alignment correction amount, and outputs the reference image. The inspection object of the comparative image 33 which is brought closer to the inspection object is formed. That is, a comparison image 33 is generated by correcting the position of the block image of the inspection image 32 in the XY directions. Specifically, the block image 44a of the comparative image 33 is an image obtained by correcting the block image 43a of the inspection image 32 (DI) by the amount of correction for alignment.

  The comparison unit 14 compares the inspection target whose position in each of the block images 44a to 44n of the comparison image 33 whose position of the inspection image 32 has been corrected and the inspection target of the reference image 31 with each pixel. The determination unit 15 calculates a difference between the brightness values of the pixels of the reference image and the pixels of the comparison image in the comparison result of the reference image and the comparison image, and determines a pixel that deviates from a preset threshold range as a defective pixel. When the number of defective pixels exceeds a preset number, a defect is determined for the inspection image. The determination result is displayed on the display unit or output to another device.

  In addition, the comparison image generation unit 13 further compares the frame 41 of the frame image 41 of the reference image 31 with respect to the partial frame of the frame 41 having the inclination in each block image of the inspection image 32 in FIG. A rotation correction unit may be provided that rotates the coordinates to correct the rotation so as to match the direction of the partial frame.

The image inspection performed by the image inspection apparatus 100 according to the present embodiment will be described with reference to the flowchart illustrated in FIG.
First, image data (reference image REM) including a character, a symbol, and a pattern to be inspected for a non-defective product, which is a comparison reference and is captured by the imaging unit 1 according to an instruction from the control unit 3, is input to the image processing unit 2 (step S1). ).

  Next, in the reference image REM, a positioning point (POS point) composed of feature points each having position information is detected by the positioning point detection unit 16 (step S2). Further, the control unit 3 creates position information (coordinate information) of the POS point as map data for the reference image, and stores it in the storage unit 6 together with the reference image REM (step S3).

  Next, the imaging unit 1 captures the characters, symbols, and patterns to be inspected to generate an inspection image (step S4). The captured inspection image is subjected to the above-described image processing by the image processing unit 2, and outputs an inspection image DI including image data in which gradations and the like are quantified to the block division unit 11. The block division unit 11 divides the inspection image DI into block images each having a preset division size (for example, 16 pixels × 128 pixels) (step S5). At this time, XY coordinate information is given to the inspection image DI. Note that the section size of the block image is an arbitrary size, and is appropriately set depending on the inspection target.

  Next, for the divided block images, the positional shift amount is detected by the positional shift amount calculation unit 21 as described above (step S6). Next, the block combining unit 22 calculates the number of POM points for which the amount of displacement has been calculated, and determines whether the calculated number of POM points is equal to or greater than a set number (step S7). If the number of detected POM points is less than the set number (NO), adjacent block images are combined so that four or more POM points are included in a group of block images (step S8). Further, the correction amount calculation unit 23 calculates a position correction amount of each block image from the position shift amount of the POM point (Step S9).

Next, each block image of the inspection image (DIB) is corrected using the alignment correction amount to form a comparison image 33 close to the block image of the reference image (step S10).
Next, the comparison image 33 is compared with the reference image 31 (REMA) based on the block image (step S11). In this comparison, each pixel of the inspection image is compared with threshold data, and a pixel that deviates from a preset threshold range is determined as a defective pixel, and the result is displayed on a display unit or output to another device (step S12). . Next, it is determined whether the quality determination of all the inspection images has been completed (step S13). If the pass / fail determination has not been completed for all the images (NO), the process returns to step S4 to capture the next inspection image and continue the inspection. When the pass / fail determination of all the images is completed (YES), the image inspection ends.

  As described above, in the image inspection apparatus according to the present embodiment, the same inspection image in the captured inspection image is inclined and misaligned with respect to the inspection target such as a character or a pattern included in the reference image. In this case, since the inspection image is divided into blocks, it is possible to calculate a correction amount for alignment in units of block images. As compared with the conventional correction of a line block image, highly accurate position correction can be realized.

  In addition, when the number of alignment points in one block image is less than the set number, the block images are combined until the set number is reached, and the alignment correction amount for correcting the positional shift is calculated, thereby It is possible to prevent the block from being corrected to a misregistered position or not being corrected at all.

  Further, in the conventional correction of the line block image, when the inspection target of the captured inspection image has an inclination with respect to the inspection target of the reference image, even if there is no pixel loss such as a character missing, the inclination is adjusted. Has been determined to be defective. However, since position adjustment is performed to eliminate positional deviation in block image units, unnecessary defect determination can be reduced.

[Modification of First Embodiment]
Next, an image inspection apparatus according to a modification of the first embodiment will be described.
In the first embodiment, the position of the inspection target in the inspection image is corrected with respect to the inspection target in the reference image, and the comparison is performed. In this modification, the position of the inspection target in the reference image is corrected with respect to the position of the inspection target in the inspection image, and the comparison is performed. In other words, the sign is not inverted in the calculation of the correction amount, and is used as the correction amount of the reference image with respect to the inspection image. The image inspection apparatus of the present modification is equivalent to the apparatus configuration of the above-described first embodiment, and the calculation of the correction amount by the correction amount calculation unit is different.

  In this modification, an alignment point is detected from a reference image including an inspection target. At this time, coordinates are set in the reference image. The reference image is divided into block images by vertical and horizontal division. For each block image, the X direction and the Y direction of the coordinates are set for the alignment point (POS point), which is a feature point set in advance on the coordinates of the reference image, and the vicinity of the area corresponding to the POS point on the inspection image. Calculate the amount of displacement for each of the directions.

A POM point is a position alignment point at which these positional deviation amounts can be calculated. If the number of POM points in one block image is less than the preset number, the block image is combined with the adjacent block images until the number exceeds the preset number. For the block images including the combined block images, the POM points of which are equal to or more than the predetermined number, the sum of the displacements is divided by the number of the POM points to obtain an average value, which is used as a registration correction amount. An image block of a comparative image composed of the reference image corrected by the alignment correction amount may be compared with a block image of the inspection image to determine whether the inspection image is good or bad.
Also in this modification, the same effects as those in the above-described first embodiment can be obtained.

[Second Embodiment]
The second embodiment will be described with reference to FIG.
The image inspection apparatus according to the present embodiment performs image inspection on three R, G, and B block images 201, 202, and 203 of one color block image. For the R, G, and B images, similarly to the first embodiment, block division, detection of an alignment point (POS point), calculation of a position shift amount, combination of adjacent block images, and correction of a position shift. Is calculated for the alignment. In the present embodiment, since the monochrome image in the first embodiment described above is simply changed to a color image, the R, G, and B block images include an inspection object that overlaps one. include. Accordingly, in the R, G, and B block images of one color block image, the same position (coordinate position) includes the same part of the same inspection target.

In the present embodiment, the R, G, and B images of the reference image and the inspection image are each divided into block images of the same size (for example, 16 pixels × 128 pixels). Thereafter, POS points of the R block image, the G block image, and the B block image of the reference image are respectively detected.
The number of POM points in one color block image is the sum of the POM points in the R, G, and B block images on the same coordinates. When the number of POM points in one color block image is four or more, the amount of registration correction is obtained by the same calculation method as in the first embodiment.

  Further, as in the first embodiment described above, when the total POM points of the R, G, and B block images in one color block image are less than four even when the POM points are summed, By combining with a color block image adjacent to the color block image, a comparative image can be formed without displacement for each color with respect to the POM point.

  Note that the present invention is not limited to the above-described embodiment and the modified examples, and can be variously modified without departing from the gist thereof. In addition, the embodiments and the modified examples may be appropriately combined and implemented, and in that case, the combined effect is obtained.

REFERENCE SIGNS LIST 1 imaging unit 2 image processing unit 3 control unit 4 display unit 5 input unit 6 storage unit 11 block division unit 12 block image correction unit 13 comparison image generation unit , 14 ... comparator, 15 ... determiner, 16 ... alignment point detector, 21 ... position shift calculator, 22 ... block combiner, 23 ... correction calculator, 31 ... reference image, 32 ... test image, 33: comparative image, 41: frame part, 42a to 42n, 43a to 43n, 44a to 44n: block image.

Claims (7)

Providing coordinate information to the reference image including the inspection target, an alignment point detection unit that detects an alignment point from the reference image,
In the block image obtained by vertically and horizontally dividing the inspection image including the inspection target, the density of the pixel at the alignment point is compared with the density of the pixel near the coordinates corresponding to the alignment point in the block image, A displacement amount calculating unit that calculates a displacement amount due to the difference,
Pixels from which the amount of displacement can be calculated are defined as points at which the amount of displacement can be calculated. If the number of points at which the amount of displacement can be calculated is less than the set number in one block image, the number of pixels until the set number is exceeded is exceeded. A block combining unit that repeats combining with an adjacent block image,
With the inspection target of the reference image as a reference position, a correction amount detection for obtaining an alignment correction amount for correcting a positional deviation of the inspection target in the inspection image having a point at which the positional deviation amount exceeding the set number can be calculated. Department and
An image inspection apparatus comprising: comparing a block image of the inspection image corrected with the alignment correction amount with the reference image to determine whether the inspection image is good or bad. Giving the coordinate information to the reference image including the inspection target, an alignment point detection unit that detects an alignment point from the reference image,
A block division unit that divides the inspection image including the inspection target into block images by vertical and horizontal division,
A displacement amount for comparing the density of the pixel at the alignment point in the block image with the density of a pixel near the coordinates corresponding to the alignment point in the block image, and calculating the displacement amount due to the difference A calculation unit,
A pixel at which the amount of displacement can be calculated is a point at which the amount of displacement can be calculated, and if the number of points at which the amount of displacement can be calculated is less than a set number, adjacent blocks are used until the set number is exceeded. A block joining section that repeats joining,
With the inspection target included in the reference image as a reference position, a position for correcting the positional deviation of the inspection target included in the block image of the inspection image having a point at which the positional deviation amount exceeding the set number can be calculated. A correction amount detection unit for obtaining an alignment correction amount;
A comparison image forming unit that forms a comparison image obtained by correcting the block image of the inspection image with the alignment correction amount by the correction amount detection unit;
A comparison unit that compares the inspection target included in the comparison image with the inspection target included in the reference image, and calculates a comparison result based on a difference between pixel values;
For a comparison result by the comparison unit, when the number of pixels that the difference deviates from a predetermined threshold range exceeds an arbitrary set number, a determination unit that performs a failure determination on the inspection image,
An image inspection apparatus comprising:   The correction amount detecting unit calculates a position correction amount for correcting a position deviation of the reference image in the block image of the inspection image with respect to the inspection target by using a point at which a position deviation amount exceeding the set number can be calculated. The image inspection apparatus according to claim 1, wherein an average position correction amount is calculated by calculating and dividing the position correction amount by a total number of points at which the position deviation amount can be calculated.   When the block images are adjacent on both sides as the engagement order, the block combining unit preferentially combines with the block image having the larger number of points where the displacement amount can be calculated, or The image inspection apparatus according to claim 1, wherein the block images are combined in any combination order in which the block images are sequentially combined from an image and an end block image of the inspection image. Giving the coordinate information to the reference image including the inspection target, an alignment point detection unit that detects an alignment point from the reference image,
The density of the pixel at the alignment point in the block image obtained by vertically and horizontally dividing the reference image is compared with the density of pixels near the coordinates corresponding to the alignment point in the inspection image including the inspection target, and the difference is calculated. A position shift amount calculation unit for calculating a position shift amount due to
Pixels from which the amount of displacement can be calculated are defined as points at which the amount of displacement can be calculated. If the number of points at which the amount of displacement can be calculated is less than the set number in one block image, the number of pixels until the set number is exceeded is exceeded. A block combining unit that repeats combining with an adjacent block image,
With the inspection target of the inspection image as a reference position, a correction amount detection for obtaining an alignment correction amount for correcting a positional deviation of the inspection target of the reference image having a point at which the positional deviation amount exceeding the set number can be calculated. Department and
An image inspection apparatus that compares the block image of the reference image corrected by the alignment correction amount with the inspection image to determine whether the inspection image is good or defective. Giving the coordinate information to the reference image including the inspection target, detecting the alignment point from the reference image,
The inspection image including the inspection target is divided into block images by vertical and horizontal division,
The density of the pixel at the alignment point is compared with the density of a pixel near the coordinates corresponding to the alignment point in the block image, and the pixel for which the amount of positional deviation can be calculated by the difference is calculated. Extract it as a completed point,
If the number of points at which the amount of displacement detected in the inspection image can be calculated is less than the set number in one block image, the combination with adjacent block images is repeated until the set number is exceeded,
With the inspection target of the reference image as a reference position, a registration correction amount for correcting the positional deviation of the inspection target for each block image of the inspection image having a point at which the positional deviation amount exceeding the set number can be calculated. Asked,
An image inspection method for comparing a block image of the inspection image corrected with the alignment correction amount and the reference image to determine whether the inspection image is good or defective. Giving the coordinate information to the reference image including the inspection target, detecting the alignment point from the reference image,
Dividing the reference image into block images by vertical and horizontal division,
The density of the pixel at the alignment point in the block image is compared with the density of a pixel near the coordinates corresponding to the alignment point in the inspection image including the inspection target, and the amount of positional deviation due to the difference is calculated. Pixels that can be extracted are extracted as points at which the amount of displacement can be calculated,
If the number of registration points detected in the reference image is less than the set number in one block image, the combination with adjacent block images is repeated until the set number is exceeded,
With the inspection target of the inspection image as a reference position, an alignment correction amount for correcting the positional deviation of the inspection target for each block image of the reference image having the alignment points exceeding the set number,
An image inspection method for comparing the block image of the reference image corrected by the alignment correction amount with the inspection image to determine whether the inspection image is good or defective.