JP2007010525A - Defect detection method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a defect detection device capable of improving detection accuracy a print defect of a specific color or an impurity, in inspection of the print defect or the impurity included in a print or food. <P>SOLUTION: An inspection surface is imaged as a color image. The inspection surface image is compared with a master image, and defect image data of a defective part wherein the concentration level difference therefrom exceeds the first tolerance set beforehand are generated. Extraction image data wherein color elements in a single or a plurality of ranges set beforehand are extracted are generated from the second inspection surface image data wherein a color space of the inspection surface image is converted into another color space, and extracted defect image data for determining whether defect image data agree with the extraction image data or not are generated, and if the size of a defect domain in the extracted defect image data is over a tolerance set beforehand, the domain is detected as a defective spot. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a defect detection method for detecting a defect in an object to be inspected, and more particularly to a defect detection method for detecting impurities of a specific color.

Conventionally, when detecting defects in a printed matter, the printed matter is photographed with a line sensor, an area image is created from the line image of the photographed printed matter, a density level of the area image of the printed matter image, and a master image corresponding to the area image If there is a part where the density level difference between the two images exceeds the allowable value, the part is determined as a defective part, and the shape and area of the defective part are analyzed, There is a defect detection method for a sheet-like printed matter that finally determines whether the defect is a defect or a light defect (see, for example, Patent Document 1).
According to this defect detection method, the size of a defective portion having a high density level (for example, black) in the master image is substantially the same as the size of a defective portion having a low density level (for example, red or yellow). Sometimes, a defective portion can be identified regardless of the density level of the master image.

Further, the paper surface state of the printed material is detected using three kinds of color filters (R (red) filter, G (green) filter, B (blue) filter), and the detected image data is stored for each color filter. The image data is converted into H (hue), S (saturation), and I (lightness), and the HSI image data to be inspected is subtracted from the H, S, and I data of the reference image. There is a printed matter inspection device that determines whether or not the subtraction value is a preset level and inspects the state of the printed matter based on the criteria of H, S, and I (see, for example, Patent Document 2). .
JP-A-6-201611 JP-A-6-246906

  Usually, in the case of foreign matters contained in paper or the like and defective printing, very small foreign matters or defective printing are usually regarded as non-defective products. However, in packages such as pharmaceuticals, even very small contaminants and poor printing may be mistaken if blood is attached, and small contaminants and defective printing that are usually good products are also considered defective products. It is discharging.

  However, in the defect detection method for the sheet-like printed matter described in Patent Document 1, the quality is determined based on the difference in density level between the master image and the image to be inspected. When the accuracy is improved so that small impurities (defective print) other than red can be detected in the same manner as red contaminants (print defects), those that are originally good are determined as defective.

  On the other hand, in the printed matter inspection apparatus described in Patent Document 2, color data on the paper surface is converted from RGB data to HSI data, and the HSI data and the HSI data of the reference image are compared to perform pass / fail judgment. By converting from data to HSI data, it was possible to use a color space that easily reflects human color sensation and to easily set the pass / fail judgment level. If the accuracy is increased so that an object (printing defect) can be detected, small impurities other than red (defective printing) are detected in the same way as red impurities (printing defects), and those that are originally good are also considered defective. I was judging.

  Thus, it is difficult to detect even a small color for a specific color and not to detect other colors.

  In addition to printed matter, in food (confectionery), the surface image varies depending on the position of surface irregularities, surface baking conditions, regular contaminants (for example, sesame in confectionery), and the like. Since it is very large, it is difficult to detect small contaminants of a specific color only by comparing with the master image.

  In view of such problems, the present invention has an object to provide a defect detection method capable of improving the detection accuracy of a specific color contaminant or print defect in inspection of a print defect or contaminant contained in a printed matter or food. And

The defect detection method according to the invention of the present application is to detect a defect of an inspection object by imaging the inspection surface of the inspection object as a color image and detecting a defect location existing on the inspection surface based on the color image. In the method, a first inspection surface image obtained by imaging the inspection surface by an imaging device is stored in a storage unit, and the density level of the area image of the first inspection surface image and the area of the master image corresponding to the area image are stored. The density levels of the images are compared, and the defect image data of the defective portion where the density level difference between the corresponding portions of both images exceeds a preset first allowable value and the first inspected surface image are captured. Creating extracted image data obtained by extracting a single or a plurality of preset color elements from the second inspected surface image data obtained by converting the color space from the one color space to the preset second color space; The defect image data The extracted defect image data is determined to determine whether or not the area image of the data matches the area image of the extracted image data corresponding to the area image, and the size of the defect area of the extracted defect image data is preset. If it is greater than or equal to the second allowable value, the defect area is detected as a defect location.
Therefore, a defect image different from the master image is created from the master image and the image of the inspection surface, and an extraction image including a color element of a specified color is created from the inspection surface image. When the size of the region that matches the extracted image is larger than a predetermined size, it is determined as defective, so that it is possible to detect a defect such as a specified color impurity or printing failure.

  The first color space may be a color space configured by color data output from the imaging device, and the second color space may be an HSL color space.

  The extracted image data may be image data obtained by extracting color elements included in a substantially spherical space centered on a point in the second color space.

  The imaging device is a line sensor that images a sheet-like printed material traveling from one side to the other, and the line image data of the surface image to be inspected sequentially output from the imaging device is sequentially stored in a storage unit. Thus, an area image of the surface to be inspected may be created inside the storage means.

  Further, if the size of the defect area of the defect image data is equal to or larger than a preset third tolerance, the defect area may be detected as a defect location.

  The defect detection method according to the present invention extracts a specific color element area after converting the color space of the image of the inspection surface image taken as a color image into another color space. It becomes easy to do. In addition, the image data and the master image, the defect image data that has determined the defective portion such as impurities and printing defects, and the extracted image data that extracted a specific color element from the surface image to be inspected, Since it is determined whether or not the defect portion of the defect image data matches the extracted image data extraction portion, an area that does not exist in the master image and a specific color element area can be detected. Then, the accuracy of detection for a specific color can be improved by determining pass / fail according to the size of the matching region.

In addition, since the first color space is a color space composed of color data output from the image pickup apparatus, a general image pickup apparatus can be used, and defects in inspection objects such as printed matter and food can be detected at low cost. can do.
In addition, since the HSL color space is used as the second color space, it is possible to perform color determination corresponding to a human color recognition sensation.

  In addition, since color elements included in a substantially spherical space centered on a point in the second color space are extracted, it is possible to designate a range of colors close to a specific color (similar colors).

  In addition, since the sheet-like printed material is imaged using the line sensor as the imaging device, for example, the quality of the printed material can be determined by mounting the imaging device on an offset printing machine. Therefore, it can be incorporated into the line of the offset printing press, and the inspection can be performed simultaneously with printing.

  Further, since the defect is determined to be defective if the size of the defect area of the defect image is equal to or larger than a preset size, the quality determination can be performed based on the size of the defective portion other than the color.

An embodiment of a defect detection method according to the present invention will be described with reference to the drawings.
In addition, although the Example shown below demonstrated when the to-be-inspected target object was a sheet-like printed matter, printed matter other than a sheet-like printed matter may be sufficient. Further, the object to be inspected may be food.

The defect detection method according to the present invention includes, for example, a defect detection system 1 as shown in FIG. 1, an imaging device 2, and a display unit 12.
The defect detection system 1 includes an inspection surface image storage unit 3, a master image data storage unit 4, a color space conversion unit 5, a defect image data generation unit 6, an extracted color element storage unit 7, an extracted image data generation unit 8, and an extraction. The inspection apparatus includes defect image data creation means 9, defect area detection means 10, and display means 11.

The imaging device 2 is a line sensor, and image data captured by the sensor is output to the defect detection system 1. When a line sensor is used as the imaging device 2, the line sensor can be mounted on an offset printing machine, and inspection can be performed simultaneously with printing of a sheet-like printed material such as a sheet.
As the imaging device 2, the line image data obtained by dividing the surface to be inspected as a line sensor may be sequentially stored in the storage unit, and an area image of the surface to be inspected may be created inside the storage unit. . Alternatively, an area image may be created by photographing the entire surface to be inspected at a time like a digital camera.
The image data output from the image pickup device 2 is created in the RGB color space when the sensor of the image pickup device 2 is composed of R (red), G (green), and B (blue). The first inspected surface image data may be output. Further, not only the RGB sensor but also a color space captured by an RGB sensor and a sensor (for example, Y (yellow)) that supplements the sensor may be used. For example, when the image is picked up by an RGBY sensor, the data output from each sensor may be stored in the storage means. However, the image pickup apparatus 2 is provided with an RGB conversion means for converting RGBY data into RGB data. You may output after converting RGBY data into RGB data.
As described above, if the color data output from the imaging device 2 is data in the RGB color space, a general imaging device can be used as the imaging device 2 so that inspection can be performed at low cost. Become.

  The inspected surface image storage means 3 is means for storing first inspected surface image data 13 (see FIG. 3) captured by the imaging device 2. The master image data storage means 4 is a means for preliminarily storing master image data 14 (see FIG. 3), which is a quality judgment criterion.

  The defect image data creating means 6 compares the density level of the area image of the first inspection surface image data 13 with the density level of the master image data 14 corresponding to the area image, and the density level difference is determined. This is means for creating defect image data 16 (see FIG. 3) of a defective portion exceeding a preset first allowable value.

The color space conversion means 5 converts the first inspection surface image data 13 in the RGB color space stored in the inspection surface image storage means 3 into H (Hue), S (Saturation), This is a means for converting the color space into the second inspected surface image data 15 (see FIG. 3) in the L (Lightness) color space, and the converted second color space is not limited to the HSL color space. May be an L ^* a ^* b ^* color space, an HSV color space, a Munsell color space (Munsell color system), or other color space (color system). If the color space is compatible, it is easy to detect colors close to a specific color, but YMCK (Y (yellow), M (magenta), C ( Cyan) and K (black)) color spaces may be converted.
The color space conversion method can be converted by a known means or a combination thereof, and therefore detailed description thereof is omitted in this embodiment.

  The extracted color element storage means 7 is a means for storing one or a plurality of color element data 17 (see FIG. 3) to be extracted. When the second color space is the HSL color space, the H, S, and L elements are stored as the extracted color element data 17 (see FIG. 3). Further, as shown in FIG. 5, the color element of the point P in the HSL color space S and the radius r centered on the point P are stored, and the spherical element having the radius r centered on the point P is stored. The included color elements may be extracted by the extracted image data creation means 8.

  The extracted image data creation means 8 includes color elements stored in the extracted color element storage means 7 from the second surface image data 15 whose color space has been converted by the color space conversion means 5. This is means for creating extracted image data 18 (see FIG. 3) from which regions have been extracted.

  Based on the area image of the defect image data 16 and the area image of the extracted image data 18, the extracted defect image data creation means 9 and the defect areas 24,. This is means for extracting extracted regions 26 and 27 and creating extracted defect image data 19 by extracting an image of a defective part including a specific color from the defective part.

The defect area detection means 10 determines whether or not the size of all the defective portions in the area image of the extracted defect image data 19 is smaller than the second allowable value, and if it is equal to or larger than the second allowable value. If it is smaller than that, it is determined as a normal location. If there is even one defective portion, it is means for determining a printed material (inspected object) as a defective product.
Further, the defective area detection means 10 determines a defective portion if each defective portion in the defect image data 16 is equal to or greater than a third tolerance value, and determines a normal location if it is equal to or smaller than the third tolerance value. And when it determines about all the defective parts, when even one defective part exists, it is a means which determines a printed matter (object to be inspected) as a defective product.

  The display unit 11 is a unit that displays an image including a defective portion on the display unit 12 when the defective area detecting unit 10 determines that there is one or more defective portions.

  Next, processing of the defect detection method will be described with reference to FIGS. FIG. 2 is a flowchart explaining the processing of the defect detection method, FIG. 3 is an explanatory diagram regarding data used in the defect detection method, and FIG. 4 is an explanation of the data shown in FIG. 3 as an area image. FIG.

  First, the master image data 14 is stored in the master image data storage unit 4. The area image 14a of the master image data 14 in this embodiment is a color image of a red rectangle 20R and a green rectangle 20G shown in FIG.

  Then, the surface to be inspected is imaged by the imaging device 2 (S1), and the first inspection surface image data 13 in the RGB color space is stored in the storage means 3 (S2). Here, as the area image 13a of the inspection surface image, as shown in FIG. 4B, a red rectangle 21R, a green rectangle 21G, a red circle 22R, blue circles 22B and 22B, yellow Color images of the round shapes 22Y and 22Y and the black round shape 23K are used. In FIGS. 4A and 4B, each color is expressed by hatching or the like. In addition, the pattern of the master image and the normal pattern printed on the printed material are simplified to rectangles 20 and 21, and the contaminants (printing defects) of the printed material are simplified to round shapes 22 and 23. The shape and size of the master image and the shape and size of impurities (printing defects) are not limited to those illustrated.

When the first inspection surface image data 13 is stored in step S2 (S2), the defect image data creating means 6 uses the master image data 14 and the first inspection surface image data 13 to calculate both images. Defect image data 16 of the defect portion where the density level difference exceeds the first allowable value is created (S3). As the area image 16a of the defect image data 16, data including round areas 24,..., 24 and a larger area 25 as shown in FIG.
As a density level comparison method, there is a method in which a master image and an inspection surface image are divided into regions of a predetermined size, and the defect levels are detected by comparing the density levels of both divided images. Further, there is a method of detecting a defective portion by comparing density levels for each pixel when the resolutions of the master image data and the first inspected surface image data are the same. Furthermore, you may detect by another well-known method.
Note that when the tolerance value of the first tolerance value is decreased, a portion where the difference between the two density levels is small is also detected, and many defective portions can be detected. On the other hand, when the allowable value is increased, a portion where the density level difference is small is not detected.

When the defect image data 16 is created, it is determined whether or not the area (area) of each defect portion is smaller than a third allowable value. If the defect area 25 in the area image 16a in FIG. 4C is larger than the third allowable value, it is determined as a defective product, and the process proceeds to Step 9 (S9) described later.
On the other hand, when the value is smaller than the third allowable value of the defect area 25, the size of the defect area is determined as non-defective, and the process proceeds to step 5 (S5).
Similarly, for the other defect areas 24,..., 24, it is determined whether or not the size of the area of each defective portion is smaller than the third allowable value, and the quality is determined.
Further, it is determined that the defect area is non-defective when the size of the defect area is smaller than the third allowable value, and when one or more locations where the size of the defect area is larger than the third allowable value exist, judge.

In step 4 (S4), when the size of the defect area is judged as good or bad and not defective, the color of the area image of the first inspection surface image data 13 in the RGB color space is converted to HSL color space, 2 Create surface image data 15 to be inspected. The area image of the second inspection surface image data 15 is the same as that shown in FIG. 4B, but is composed of H (hue), S (saturation), and L (lightness) shown in FIG. This is data of the color space S.
This HSL color space is a color space that is closer to the human color sense than the RGB color space. Therefore, as shown in FIG. 5, the spherical space within the radius r centered on one point P in this space is a color very close to the color of the point P as a human sense. By specifying a position (point P) in the HSL space and a distance (radius r) centered on the position, an area image of a specific color can be extracted from the image data of the printed matter. .
FIG. 5 illustrates a spherical shape with a radius r centered on the point P. However, for example, when it is desired to widen the extraction range related to lightness, the lightness is included in an elliptical spherical space whose major axis is the lightness axis direction. Color elements may be extracted.

Then, extracted image data 18 is created from the second inspection surface image data 15 whose data has been converted into the HSL color space and the extracted color element data 17 (S6). When the color elements of the red rectangle 21R and the red circle 22R of the second inspection surface image data 15 are included in the color elements of the extracted color element data 17, as shown in FIG. A rectangular extraction area 26 corresponding to the red rectangle 21 and a circular extraction area 27 corresponding to the red circle 22R are extracted.
As a method for creating the extracted image data, the image of the second inspection surface image data is divided into a predetermined size, and the specified color element is included in this divided image, as in the above-described density level comparison method. A method for detecting an area including the specified color element, and whether or not the specified color element is included for each pixel of the image of the second inspection surface image data. There is a method for detecting an area including a color element.

Then, from the defect image data 16 and the extracted image data 18, the area image 16 a of the defect image data is compared with the area image 18 a of the extracted image data corresponding to the area image 16 a to extract a matching area. Defect image data is created (S7). That is, when the area image 16a in FIG. 4C is compared with the area image 18a in FIG. 4D, the circular area 24r located in the upper left of FIG. 4C and the circular extraction area 27 in FIG. However, since the other defect areas 24,..., 24, 25 do not match the extraction areas 26, 27 of the extracted image data, an area image 19a as shown in FIG. .
As a comparison method between the area image 16a of the defect image data and the area image 18a of the extracted image data, both of the area images are divided into predetermined sizes as in the above-described comparison method of the density level difference. There are a method of comparing corresponding divided images, a method of comparing each corresponding pixel, and the like.

Then, it is determined whether or not the size of the extracted defect area 28 in the area image 19a of the extracted defect image data is smaller than the second allowable value. If it is smaller than the second allowable value, the printed matter is determined as a non-defective product, and if it is larger than the second allowable value, it is determined as a defective product (S8).
When measuring the size of the extracted defect area 28 in the area image 19a of the extracted defect image data, the horizontal length X and the vertical Y length of the extracted defect area 28 are shown in FIG. The size is measured, and the size is obtained from X and Y.
It should be noted that when there are a plurality of extracted defect areas, measurement is performed for each area, and when at least one area is larger than the second allowable value, it is determined as a defective product.

  When a defective part is detected in step 4 (S4) and step 8 (S8), an area image (part of the area image) including the defective part is displayed on the display unit 12 (S9). In this way, by displaying the area image of the defective portion on the display unit 12, the defective portion can be identified.

In the above-described embodiment, the extracted color element data 17 has been described by taking red as an example. However, as a matter of course, other colors or a plurality of colors may be extracted.
Note that even when the object to be inspected is food, a master image serving as a reference is stored in advance, and a foreign object contained in the food can be detected by imaging the food to be inspected.

It is a block diagram regarding a defect detection system. It is a flowchart explaining the process of the defect detection method. It is explanatory drawing of the data used with a defect detection method. It is explanatory drawing of the area image of each data in a defect detection method. It is explanatory drawing of HSL color space.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Defect detection system 2 Imaging device 3 Inspection surface image storage means 4 Master image data storage means 5 Color space conversion means 6 Defect image data creation means 7 Extracted color element storage means 8 Extracted image data creation means 9 Extracted defect image data creation means DESCRIPTION OF SYMBOLS 10 Defect area | region detection means 11 Display means 12 Display part 13 1st to-be-inspected surface image data 14 Master image data 15 2nd to-be-inspected surface image data 16 Defect image data 17 Extracted color element data 18 Extracted image data 19 Extracted defect image data 20 , 21 Normal pattern 22, 23 Defective part 24, 25 Defect region 26, 27 Extraction region 28 Extraction defect region P Center point r Radius S HSL color space

Claims (5)

In the defect detection method for an inspection target object obtained by capturing the inspection surface of the inspection target object as a color image and detecting a defect location existing on the inspection surface based on the color image.
Storing a first inspection surface image obtained by imaging the inspection surface by an imaging device in a storage unit;
The density level of the area image of the first surface image to be inspected is compared with the density level of the area image of the master image corresponding to the area image, and the density level difference between the corresponding portions of both images is preset. From the defect image data of the defect portion exceeding the allowable value and the second inspection surface image data obtained by converting the color space from the first color space obtained by imaging the first inspection surface image to the preset second color space, Creating extracted image data obtained by extracting color elements of a single or a plurality of preset ranges,
Create extracted defect image data that determines whether or not the area image of the defect image data and the area image of the extracted image data corresponding to the area image match,
A defect detection method comprising detecting the defect area as a defect location if the size of the defect area of the extracted defect image data is greater than or equal to a preset second tolerance. The first color space is a color space constituted by color data output from the imaging device,
The defect detection method according to claim 1, wherein the second color space is an HSL color space.   The defect detection method according to claim 1, wherein the extracted image data is image data obtained by extracting color elements included in a substantially spherical space centered on a point in the second color space. The imaging device is a line sensor that images a sheet-like printed matter that travels from one side to the other,
The line image data of the surface image to be inspected that is sequentially output from the imaging device is sequentially stored in a storage unit, and an area image of the surface to be inspected is created in the storage unit. Defect detection method. The defect detection method according to claim 1, wherein the defect area is detected as a defect location if the size of the missing area of the defect image data is greater than or equal to a preset third tolerance.

Images