JP5074998B2 - Appearance inspection method and apparatus for transparent film - Google Patents

Description

  The present invention relates to a transparent film visual inspection method and a transparent film visual inspection apparatus for performing visual inspection for the presence or absence of defects such as foreign matter and bubbles by performing computer image processing on an image of a transparent film having low reflectance. is there.

  Generally, a transparent film formed of a synthetic resin material becomes a defective product when foreign matter or bubbles are mixed therein, and therefore, inspection items of the transparent film include the presence or absence of defects such as foreign matter and bubbles. The inspection for the presence or absence of defects is performed by irradiating light to a transparent film to be inspected and detecting transmitted light or reflected light. However, when the inspector visually inspects, it is considered that the inspection results vary depending on the inspection ability and physical condition, and the number of oversights increases.

  In order to solve this type of problem, it is possible to perform a visual inspection for the presence or absence of defects by receiving transmitted light or reflected light of the light irradiated on the inspection object with a line sensor camera and using a computer image processing technique. It is considered (for example, refer to Patent Document 1). If the structure of patent document 1 is employ | adopted, the dispersion | variation in a test result will be reduced compared with the test | inspection by visual observation.

  By the way, in the technique described in Patent Document 1, a transparent film to be inspected is advanced in one direction, and the surface of the transparent film is imaged by a line sensor camera in which a plurality of CCD sensors are arranged in a direction orthogonal to the traveling direction. ing.

  As for the illumination, it is basically based on the fact that the transparent film is illuminated from the opposite side to the line sensor camera and imaged by transmitted light. Patent Document 1 also describes that a transparent film is illuminated from the same side as the line sensor camera to perform imaging with reflected light, or to perform imaging with both transmitted light and reflected light. .

  When the object to be inspected is a transparent film, the transmittance is high and the reflectance is low. Therefore, it is possible to detect the presence or absence of a defect such as an opaque foreign object by using transmitted light. The difference in received light intensity due to the presence or absence of light scattering is detected regardless of whether light is reflected or not, and therefore the change in the amount of light due to the presence or absence of defects is small, and the signal-to-noise ratio (S / N) Cannot be increased.

  In addition, it cannot be said that the inspection object has a small thickness dimension and high flatness, and even if the inspection object is a non-defective product, the received light intensity received by the line camera sensor may vary depending on the location of the inspection object. There is a possibility that the fluctuation of the received light intensity is mistaken as a defect. Further, the change in the intensity of the ambient light in the inspection environment may affect the inspection result, and there is a problem that the reproducibility of the defect detection result is low.

When the inspection object is a transparent film, the problems of defect detection described above are listed as follows.
1. Since the S / N is small, the adjustment range of the defect determination condition is narrow. That is, if the defect determination conditions are set strictly, defects are overlooked, and if the determination conditions are set loosely, misperceptions of non-defect areas as defects increase.
2. If the ambient light fluctuates, defects cannot be detected with good reproducibility.
3. There is a possibility that a local fluctuation in received light intensity is mistaken as a defect.

The technique described in Patent Document 1 employs a technique of subtracting the average value of the received light intensity (corresponding signal value) of all pixels from the received light intensity (corresponding signal value) of each pixel. It seems that problems 1 and 2 can be solved.
JP 2001-91470 A

  However, in the technique described in Patent Document 1, the average value of the light reception intensity of all the pixels is uniformly subtracted from the light reception intensity of all the pixels. Point 3 is not solved.

  The present invention has been made in view of the above-mentioned reasons, and the purpose thereof is that even if the difference in received light intensity due to the presence or absence of a defect is small, the determination condition of the defect can be set appropriately and the ambient light varies. In addition to making it possible to detect defects with high reproducibility, it is possible to accurately detect defects without misidentifying local variations in received light intensity. It is another object of the present invention to provide a transparent film appearance inspection apparatus.

  The invention of claim 1 irradiates the surface of the inspection target from the illumination device with the transparent film as the inspection target, and generates a grayscale image by receiving reflected light from each part of the surface of the inspection target by the image input device. A second process for dividing the gray image into a plurality of divided images, a third process for obtaining a mode value of the gray value in the divided image, and a non-defective range including the mode value. A fourth process of extracting pixels having gray values that deviate from the non-defective range as defect candidate points, a fifth process of integrating as a defect area a connectable area consisting of defect candidate points, and a centroid for the defect area The difference between the sixth process, the seventh process for setting the inspection area based on the center of gravity, the eighth process for determining the mode value of the gray value in the inspection area as the reference value, and the reference value is defined. Has a gray value greater than or equal to the threshold value And having a ninth step of extracting a pixel as a pixel defect.

  According to a second aspect of the invention, in the first aspect of the invention, the divided image has a number of pixels that is twice or more the number of pixels included in a defect to be extracted in the direction in which the grayscale image is divided. And

  According to a third aspect of the present invention, in the first or second aspect of the present invention, in the third step, the gray value is determined for pixels arranged on each straight line in both directions of a direction in which the gray image is divided and a direction orthogonal to the direction. In the fourth process, the logical sum of the defect candidate points obtained for each direction is adopted as the defect candidate point.

  According to a fourth aspect of the present invention, in the first or second aspect of the present invention, in the third step, the gray value is determined for pixels arranged on each straight line in both directions of a direction in which the gray image is divided and a direction orthogonal to the direction. The fourth step is characterized in that, in the fourth process, when one of the connected regions made up of defect candidate points obtained in each direction exceeds a prescribed determination area, the connected region is adopted.

  The invention of claim 5 is the invention according to any one of claims 1 to 4, wherein, in the fifth step, a connection region whose area is less than a specified minimum area is excluded from the defect regions among the connection regions composed of defect candidate points. It is characterized by doing.

  The invention of claim 6 is the invention according to any one of claims 1 to 5, wherein, in the fifth step, a connection region whose area exceeds a specified maximum area is excluded from the defect regions among the connection regions composed of defect candidate points. It is characterized by that.

  According to a seventh aspect of the present invention, an illumination device that irradiates light on the surface of the inspection target with a transparent film as an inspection target, and receives reflected light from each part of the surface of the inspection target and outputs a light reception signal corresponding to the amount of received light An image input device, an image dividing unit that divides the gray image obtained by the image input device into a plurality of divided images, and a gray value of pixels arranged on a straight line extending in a direction in which the gray image is divided in the divided image A mode value extraction unit for obtaining the mode value, and a defect candidate extraction for setting a non-defective range including the mode value obtained by the mode value extraction unit and extracting pixels having gray values that deviate from the non-defective range as defect candidate points. A defect area integration unit that integrates as a defect area a connectable area consisting of defect candidate points, and an inspection area setting unit that sets an inspection area based on the center of gravity for the defect area; Inspection A reference value setting unit that obtains a mode value of a gray value within a region as a reference value, and a defect determination unit that uses a pixel having a gray value that has a difference from the reference value equal to or greater than a predetermined threshold as a defective pixel. Features.

  According to the first and seventh aspects of the invention, the gray image is divided into divided images, the non-defective range is set so as to include the mode value of the gray value obtained in the divided image, and the gray value deviating from the non-defective range is set. Since the pixel having the pixel is extracted as a defect candidate point, the influence of color unevenness and illumination unevenness can be suppressed by limiting the range in which the defect candidate point is extracted. In other words, the non-defective product range can be set appropriately even if the difference in received light intensity due to the presence or absence of defects is small. Furthermore, since an inspection area is set for a defect area obtained by integrating connected areas including defect candidate points, and defects are extracted within the inspection area, verification of defects is limited to a narrow range including the defect area. Therefore, it becomes possible to detect the defect with high accuracy without being affected by the fluctuation of the received light intensity that occurs locally.

  According to the invention of claim 2, since the gray value of the pixel in the non-defective part becomes the mode value, it becomes possible to extract the defect candidate point with good reproducibility.

  According to the invention of claim 3, since the logical sum of the defect candidate points obtained in two directions is adopted as the defect candidate point, the number of defect candidate points can be increased, the amount of information about the defect increases, and the defect area Thus, even a minute defect can be detected regardless of the direction of the defect.

  According to the fourth aspect of the present invention, it is possible to extract a connected region including defect candidate points without depending on the direction of the defect in the grayscale image. In addition, if a condition for one direction of each direction is satisfied, the defect candidate point is used as a connection area. Therefore, if the connection area obtained in one direction satisfies the condition, there is no need to obtain a connection area in the other direction. There is a possibility that the processing time can be shortened compared to the method of Item 3.

  According to the invention of claim 5, since only the connected areas having the minimum area or more are integrated, it is possible to remove noise, and it is possible to reduce the processing time because unnecessary processing is not performed. is there.

  According to the sixth aspect of the present invention, since only the connected areas having the maximum area or less are integrated, it is possible to prevent the connected area caused by uneven color or uneven illumination from being mistaken as a defect.

  In the present embodiment, as shown in FIG. 1, an illumination device 2 that irradiates light on the surface of a transparent film 1 to be inspected, and a reflected light of the light emitted from the illumination device 2 to the surface of the transparent film 1 are received. A line sensor camera 3 as an image input device, and an image processing device 4 that determines the presence of foreign matter and bubbles in the transparent film 1 from the output of the line sensor camera 3.

  The transparent film 1 is sent in the longitudinal direction by a conveying device (not shown). Moreover, the illuminating device 2 is provided with a linear light source such as a straight tube fluorescent lamp, and has an irradiation region extending over the entire length of the transparent film 1 in the width direction (direction orthogonal to the feeding direction).

  The line sensor camera 3 uses a combination of a CCD linear image sensor and a light receiving lens made up of a cylindrical lens. The positional relationship between the illuminating device 2 and the line sensor camera 3 is set so that the regular reflection light of the light irradiated on the surface of the transparent film 1 from the illuminating device 2 enters the line sensor camera 3.

  In this embodiment, since the transparent film 1 is sent in one direction, information in the feeding direction of the transparent film 1 can be obtained by receiving the reflected light from the linear region with the line sensor camera 3. .

  The image processing apparatus 4 has a computer as a main component and executes a program that enables the following operations. The image processing device 4 includes an A / D converter 11 that converts a light reception output of the line sensor camera 3 into a digital signal, an image memory 12 that stores a gray value for each pixel output from the A / D converter 11, and Is provided. The image memory 12 stores each line of grayscale images output from the line sensor camera 3 for a plurality of lines. That is, an image (an image corresponding to a two-dimensional image) of an area having a full width in the width direction of the transparent film 1 and a predetermined length in the feed direction of the transparent film 1 is stored in the image memory 12.

  The grayscale image stored in the image memory 12 is divided into a plurality of lines in the line direction by the image dividing unit 13. Now, the direction of the arrow Y in FIG. 2A is the feeding direction of the transparent film 1, and the direction of the arrow X is the width direction of the transparent film 1 (the X direction is the direction of the line in the grayscale image and is a two-dimensional image. The Y direction corresponds to the vertical direction of the two-dimensional image). The grayscale image stored in the image memory 12 is a grayscale image P1 obtained by imaging the transparent film 1 m times by the line sensor camera 3. As shown in FIG. 2B, the image dividing unit 13 divides the grayscale image P1 into n pieces in the direction of the arrow Y (the width direction of the transparent film 1). Hereinafter, the image divided by the image dividing unit 13 is referred to as a divided image.

  If the number of pixels in the width direction of the transparent film 1 is p, the divided image P2 has m pixels in the direction of the arrow X and (p) in the direction of the arrow Y as shown in FIG. / N) having pixels. That is, the divided image P2 is a grayscale image with pm / n pixels.

  Next, the mode value extraction unit 14 determines the mode value for the gray value for each line for each divided image P2. That is, m mode values are obtained for each divided image P2. Here, the distribution of the gray value is not guaranteed to be a single peak type, but in the case of the transparent film 1, it is considered that the gray value of the part having no defect is often the mode value. Distribution. Further, the number of pixels per line (p / n) in the divided image P2 is set as appropriate, and is set to include, for example, several tens of pixels. Here, the division number n of the divided image is set so that the pixel number p of the line sensor camera 3 is divisible.

  When the mode value is obtained by the mode value extraction unit 14, the defect candidate extraction unit 15 sets a non-defective product range for each line of the grayscale image P1, and if there is a pixel that deviates from the non-defective product range, the defect candidate point is determined. . The non-defective product range is defined using the mode value for each line obtained by the mode value extraction unit 14. For example, if the mode value obtained for the kth line is A (k), the upper limit of the non-defective product range is set to A (k) + B, and the lower limit of the non-defective product range is set to A (k) -C. The values B and C may be equal or different.

  That is, since the gray value of the part where the defect exists is usually larger or smaller than the mode value (the gray value of the non-defect portion), the non-defective range is defined as described above, and the gray value is larger than the non-defective range. A large or small region is extracted as a defect candidate point.

  Now, it is assumed that the defect candidate points extracted by the defect candidate extraction unit 15 are distributed as shown in FIG. If the defect candidate point actually corresponds to the defect, the defect candidate point forms a connected region D. In order to form the connected region D, when the pixels that are out of the non-defective range are included in the pixels in the vicinity of the pixels that deviate from the non-defective range, an operation that handles both pixels as one group of pixels is performed. A pixel group obtained by repetition is defined as a pixel group in the connection region.

  Since the defect candidate extraction unit 15 extracts defect candidate points for each divided image P2, the extracted defect candidate points are arranged in the gray image P1 before division to form the above-described defect candidate point connection region.

  Here, the divided image P2 obtained by dividing the grayscale image P1 is less affected by color unevenness and illuminance unevenness as the width is narrowed, but is included in the line when obtaining the mode value of the grayscale value for each line. If the number of pixels is smaller than the number of defect candidate points, the mode value cannot be obtained as the gray value of the non-defect portion. In practice, it is necessary to set the mode value to at least twice the number of pixels at the defect candidate point. Since the pixel number of the defect candidate point can be obtained experimentally according to the specifications of the apparatus, the maximum value of the pixel number of the defect candidate point per line in the divided image P2 is obtained, and more than twice that value. The number of pixels of the divided image P2 is determined so that the number of pixels is included.

  By the way, since the ambient light in the inspection environment may change in a relatively short time, it is necessary to obtain the received light intensity at a timing that is not affected by the change in the ambient light for the pixel used when obtaining the mode value. . In this embodiment, the line sensor camera 3 is used to image the transparent film 1, and the line sensor camera 3 can acquire an image of one line in a time of 1/1000 second or less. With respect to the minute pixels, it is possible to acquire the received light intensity at a timing that can be said to be substantially simultaneous. However, it is not essential to use a CCD linear image sensor as the line sensor camera 3, and other linear image sensors can be used.

  In an inspection environment in which ambient light hardly changes, the mode value may be obtained from a pixel row arranged in a direction orthogonal to the line. Further, defect candidate points may be obtained for both the pixel column on the line and the pixel column in the direction orthogonal to the line, and the defect candidate point may be determined by combining both results. When combining both results as defect candidate points, the logical sum of the defect candidate points obtained in both directions can be used. Alternatively, defect candidate points are obtained for both the pixel row on the line and the pixel row in the direction orthogonal to the line, and one area of each connected region composed of the defect candidate points in each direction is greater than or equal to the determined determination area. For example, the connected region is adopted as the output of the defect candidate extraction unit 15.

  After the connection area of defect candidate points is formed, the defect area integration unit 16 evaluates whether or not the connection areas of the defect candidate points can be integrated into one group, and if integration is possible, groups them as defect areas. . In the evaluation of whether or not integration is possible, the distance between the defect candidate points is obtained, and if the distance is within the specified integration distance, it is determined that integration is possible.

  For example, it is assumed that connection regions A1, A2, A3, B1, B2, and B3 of defect candidate points are obtained as shown in FIG. Here, it is assumed that the connection regions A1, A2, and A3 are one defect. That is, in the illustrated example, a region corresponding to one defect is separated into a plurality (three in the illustrated example) of connected regions A1, A2, and A3. For the connection areas A1, A2, and A3 thus separated, the defect area integration unit 16 calculates the distances between the defect candidate points included in the different connection areas A1, A2, and A3, and the calculated distances are defined. If it is within the integrated distance, it is integrated as one defective area.

  On the other hand, when the area of the connection region including defect candidate points is extremely small (corresponding to the region B2 in FIG. 4), there is a possibility of noise in the region. Therefore, the area of the connection region composed of defect candidate points is evaluated, and the connection region that does not satisfy the specified minimum area (number of pixels) is not adopted as the defect region. On the other hand, when the area of the connection area composed of defect candidate points is large (corresponding to the area B3 in FIG. 4), the area is not a defect but may be color unevenness or illumination unevenness. Of these, those exceeding the specified maximum area (number of pixels) are not adopted as defect areas. That is, each of the connection areas A1, A2, A3, B1, B2, B3 is integrated as a defect area if the area is within a specified range and the distance between the defect candidate points is within the integration distance. .

  For the defect region integrated by the defect region integration unit 16, the center of gravity calculation unit 17 obtains the center of gravity position. After obtaining the center of gravity position, the inspection region setting unit 18 sets an inspection region with the center of gravity as a reference point. The shape of the inspection area is not particularly limited, but a rectangular inspection area is usually set. After limiting the inspection area in this way, the reference value setting unit 19 obtains the mode value of the gray value for the pixels in the inspection area and sets it as the reference value. In the inspection area, it is considered that the gray value of the area having no defect becomes the mode value. Therefore, the mode value is set as the reference value, and the defect determination unit 20 defines the difference from the reference value among the pixels in the inspection area. When pixels having a gray value that is equal to or greater than the threshold value form a connected region, and the connected region includes pixels that are equal to or greater than the prescribed number of pixels, the connected region is determined to be a region corresponding to a defect. If there is a pixel recognized as a defect by the defect determination unit 20, it is determined that there is a defect such as a foreign object or a bubble in the transparent film 1, and the position of the defect is specified from the position of the pixel.

  The processing procedure described above is summarized in FIG. First, the density including pixels for a plurality of lines is obtained by imaging the transparent film 1 conveyed in one direction a plurality of times by the line sensor camera 3 in which pixels are arranged in a direction orthogonal to the conveyance direction (width direction of the transparent film 1). An image P1 is acquired (S1). Next, the acquired gray image P1 is divided into a plurality of divided images P2 in the line direction (S2). Thereafter, the gray value of each pixel for each line in each divided image P2 is obtained, and the mode value of the gray value is obtained (S3). A non-defective range is defined for each line on the basis of the mode value, and pixels that deviate from the non-defective range are obtained as defect candidate points (S4). Since the defect candidate points form a connection area, the connection area is evaluated, and whether or not the connection areas can be integrated is determined, and the connection areas that can be integrated are integrated to determine the defect area (S5).

  When the defect area is determined, the center of gravity of the defect area is calculated (S6), and the inspection area is set using the center of gravity as a reference point (S7). The mode value of the gray value is obtained in the inspection area (S8), and the connected area including the pixels whose gray level difference from the obtained mode value is equal to or larger than the defined threshold value is included in the inspection area. Sometimes it is finally recognized as a defect (S9).

  The presence or absence of a defect in the transparent film 1 can be inspected by the above processing procedure. In this inspection, by using the divided image P2 and obtaining the mode value for each line, the influence of uneven color and uneven illumination can be suppressed. In addition, since the defect candidate points are extracted by setting the non-defective range based on the mode value for each line, there is a high possibility that an appropriate non-defective range can be set even if the difference in received light intensity due to the presence or absence of defects is small. Become. Furthermore, since the inspection area is set in a narrow range including the defect area after the defect area is extracted, it becomes possible to detect the defect accurately without misidentifying the fluctuation of the received light intensity that occurs locally. .

It is a schematic structure figure showing an embodiment of the present invention. It is operation | movement explanatory drawing same as the above. It is operation | movement explanatory drawing same as the above. It is operation | movement explanatory drawing same as the above. It is operation | movement explanatory drawing which shows the process sequence same as the above.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Transparent film 2 Illuminating device 3 Line sensor camera 4 Image processing apparatus 11 A / D converter 12 Image memory 13 Image division part 14 Mode value extraction part 15 Defect candidate extraction part 16 Defect area integration part 17 Center of gravity calculation part 18 Inspection area Setting unit 19 Reference value setting unit 20 Defect determination unit P1 Gray image P2 Divided image

Claims (7)

  A first process of irradiating the surface of the inspection object from the illumination device with the transparent film as the inspection object, and receiving a reflected light from each part of the surface of the inspection object by the image input device, and generating a grayscale image; A second step of dividing the gray image into a plurality of divided images; a third step of obtaining a mode value of the gray value in the divided image; and a gray value that sets a non-defective range including the mode value and deviates from the non-defective range A fourth process for extracting pixels having a defect candidate point, a fifth process for integrating a connectable area composed of defect candidate points as a defect area, a sixth process for obtaining a center of gravity for the defect area, A seventh process for setting an inspection area based on the center of gravity, an eighth process for obtaining a mode value of a gray value within the inspection area as a reference value, and a gray value at which a difference between the reference value is equal to or greater than a predetermined threshold value Pixels with defective pixels Appearance inspection method for a transparent film and having a ninth step of to extract.   2. The transparent film appearance inspection method according to claim 1, wherein the divided image has a number of pixels equal to or greater than twice the number of pixels included in a defect to be extracted in a direction in which the grayscale image is divided.   In the third process, the mode value of the gray value is obtained for each pixel arranged on each straight line in both the direction in which the gray image is divided and the direction orthogonal to the direction, and in the fourth process, the mode value is obtained for each direction. 3. The method for inspecting the appearance of a transparent film according to claim 1, wherein a logical sum of defect candidate points is adopted as a defect candidate point.   In the third process, the mode value of the gray value is obtained for each pixel arranged on each straight line in both the direction in which the gray image is divided and the direction orthogonal to the direction, and in the fourth process, the mode value is obtained for each direction. 3. The transparent film appearance inspection method according to claim 1 or 2, wherein one of the connection areas composed of defect candidate points exceeds the prescribed determination area.   5. The method according to claim 1, wherein, in the fifth step, a connection region whose area is less than a specified minimum area is excluded from the defect regions among the connection regions including defect candidate points. Appearance inspection method of transparent film.   6. The transparent according to claim 1, wherein, in the fifth step, a connected region whose area exceeds a specified maximum area is excluded from the defective regions among the connected regions including defect candidate points. Film appearance inspection method.   An illumination device that irradiates light onto the surface of the inspection target using a transparent film as an inspection target, an image input device that receives reflected light from each part of the surface of the inspection target and outputs a received light signal according to the amount of received light, and image input An image dividing unit that divides a gray image obtained by the apparatus into a plurality of divided images, and a mode for obtaining a mode value of gray values for pixels arranged on a straight line extending in a direction in which the gray image is divided in the divided image. From the defect candidate point and the defect candidate point, a defect candidate extraction unit that sets a non-defective product range including the mode value obtained by the mode extraction unit, and extracts pixels having gray values that deviate from the good product range as defect candidate points A defect area integration unit that integrates the connectable areas that can be integrated as a defect area, an inspection area setting unit that obtains a center of gravity for the defect area and sets an inspection area based on the center of gravity, and a gray value in the inspection area Most An external appearance of a transparent film, comprising: a reference value setting unit that obtains a value as a reference value; and a defect determination unit that uses a pixel having a gray value whose difference from the reference value is equal to or greater than a specified threshold as a defective pixel Inspection device.

Images