TWI607211B - Image generation device, defect inspection device, and defect inspection method - Google Patents

Description

Image generating device, defect inspection device, and defect inspection method

The present invention relates to an image generating apparatus that generates image data for inspecting a defect of a sheet-like molded body such as a polarizing film or a retardation film, a defect inspecting apparatus including the image generating apparatus, and a defect inspecting method.

Previously, the defect inspection device examined defects of a sheet-shaped formed body such as a polarizing film or a retardation film using a one-dimensional camera called a line sensor. In the state in which the sheet-shaped molded body is illuminated by a linear light source such as a fluorescent tube, the surface of the sheet-shaped molded body is scanned from one end of the longitudinal direction along the longitudinal direction of the sheet-shaped formed body by a line sensor. To the other end, one or more pieces of one-dimensional image data (still image data) are acquired. Then, two-dimensional image data is generated by closely framing a plurality of one-dimensional image data in order of acquisition time, and the defects of the sheet-shaped formed body are inspected based on the two-dimensional image data.

One-dimensional image data acquired by the line sensor usually includes a linear light source image. When the linear light source and the line sensor are disposed on one side of the sheet-shaped molded body, the linear light source image is emitted from the linear light source and is unidirectionally reflected by the sheet-shaped formed body to reach the light of the line sensor. Image. Further, when the sheet-like molded body is disposed between the linear light source and the line sensor, the linear light source image is emitted from the linear light source and transmitted through the sheet-shaped molded body to reach the light of the line sensor. . In the case of the defect inspection apparatus, when the width of the sheet-like formed body is wide, the plurality of line sensors are arranged in the width direction so that the entire area in the width direction of the sheet-shaped formed body can be inspected.

In the prior defect inspection device, due to the use of a plurality of one-dimensional images The two-dimensional image data generated by the close-packing is used to inspect the defects of the sheet-shaped molded body, so that the positional relationship between the inspection target pixel and the linear light source image in each of the one-dimensional image data constituting the two-dimensional image data becomes one Determine the positional relationship. The defect exists only when the positional relationship between the inspection target pixel (the target pixel) and the linear light source image is in a specific positional relationship, and appears on the one-dimensional image data. For example, a bubble which is a defect is often present on a one-dimensional image data only when it is located at or near the periphery of the image of the linear light source. As such, defects are sometimes not detected due to their location. Therefore, the above-described defect inspection apparatus for inspecting the defects of the sheet-shaped formed body using the two-dimensional image data composed of the plurality of one-dimensional image data acquired by the line sensor has only limited defect detecting capability.

As a defect inspecting device which solves the above-mentioned problem, a device is disclosed in Japanese Patent Laid-Open Publication No. 2007-218629 (Patent Document 1) and Japanese Patent Laid-Open No. 2010-122192 (Patent Document 2). The device illuminates the sheet-shaped molded body with a linear light source such as a fluorescent tube, and continuously conveys the sheet-shaped molded body in a specific transport direction, and acquires two-dimensional image data using a two-dimensional camera called a region sensor ( The animation data), and the defects of the sheet-shaped formed body are inspected based on the two-dimensional image data.

According to the defect inspection apparatus disclosed in Patent Documents 1 and 2, since it is possible to determine whether or not there is a defect based on a plurality of pieces of two-dimensional image data having different positional relationships between the inspection target pixel and the linear light source image, the previous use of the line sensor is used. The defect can be detected more reliably than the defect inspection device. Therefore, the defect inspection device using the area sensor disclosed in Patent Documents 1 and 2 has an improved defect detection capability as compared with the previous defect inspection device using the line sensor.

The defect inspection device using the area sensor disclosed in Patent Documents 1 and 2 must process two-dimensional image data having a large amount of information. In the defect inspection device, the two-dimensional image data output from the area sensor is used for the personal computer (PC, Personal) In the image analysis unit realized by Computer, the defect position and the like are analyzed. However, since the amount of information of the two-dimensional image data is large, the analysis processing time of the two-dimensional image data by the image analysis unit becomes long. tendency.

In the defect inspection device, the conveyance speed of the sheet-shaped formed body is controlled based on the analysis processing speed of the two-dimensional image data by the image analysis unit. When the analysis processing speed of the image analysis unit using the two-dimensional image data having a large amount of information is slow, the conveyance speed of the sheet-shaped formed body must be lowered, and the inspection efficiency is lowered.

An object of the present invention is to provide an image generating apparatus, a defect inspecting apparatus including the image generating apparatus, and a defect inspecting method, the image generating apparatus generating image data for inspecting defects of the sheet-shaped formed body, and Before the high defect detection capability is maintained, the image processing by the image analysis unit can be speeded up, and the inspection efficiency can be improved.

The present invention provides an image generating apparatus that generates image data for inspecting a defect of a sheet-shaped formed body, and includes: a conveying portion that conveys the sheet-shaped formed body along a length direction of the sheet-shaped formed body; a portion including a light source extending linearly in a width direction perpendicular to the longitudinal direction of the sheet-shaped formed body, and the sheet-shaped molded body is irradiated with light by the light source; and the imaging unit is configured to transport the sheet conveyed by the transport unit The shaped body performs an imaging operation to generate two-dimensional image data representing a two-dimensional image; the feature value calculating unit processes one or more kinds of algorithms, and calculates the two-dimensional map based on the luminance values of the respective pixels. a processed image data generating unit that divides each pixel constituting the two-dimensional image data into a defective pixel whose feature value is equal to or greater than a predetermined threshold value, and the feature value does not reach the above a remaining pixel of the threshold, and generating processed image data, the processed image data comprising, for the defective pixel, storing grayscale information indicating a grayscale value corresponding to the feature value The grayscale information stores a bit string and includes a representation zero for the remaining pixels a gray scale information storage bit string of the gray scale information of the gray scale value; the defect information acquisition unit obtains defect information about the defect in the sheet shaped body for each pixel according to the processed image data, and generates the storage a defect information storage bit string having the defect information obtained as described above; and an analysis image data generating unit that generates analysis image data for each pixel, the analysis image data including the processed image data The grayscale information storage bit string is added to the parsing bit string obtained by the defect information storage bit string.

Further, in the image generating apparatus of the present invention, the defect information may include defect type information indicating a type of a defect in the sheet-shaped formed body.

Further, in the image generating apparatus of the present invention, preferably, the feature value calculating unit calculates the feature value by a plurality of types of algorithm processing, and the defect information acquiring unit is based on the gray scale information of each pixel. The grayscale information of the stored bit string is the grayscale information corresponding to the feature value calculated by the processing of any one of the plurality of algorithm processes, and the defect information including the defect type information is obtained.

Moreover, the present invention provides a defect inspection apparatus including: the image generation device; and an image analysis device using an analysis map generated by the analysis image data generation unit constituting the image generation device. The image analysis is performed using the information of the bit string, and predetermined image analysis is performed to detect defects of the sheet-shaped formed body.

Moreover, the present invention is a defect inspection method for inspecting a defect of a sheet-shaped formed body, and includes: a conveying step of conveying the sheet-shaped formed body along a length direction of the sheet-shaped formed body; and a light irradiation step a light source that linearly extends in a width direction perpendicular to the longitudinal direction of the sheet-shaped molded body, irradiates the sheet-shaped molded body to be conveyed with light, and an imaging step of performing an image capturing operation on the sheet-shaped formed body during conveyance. produce a two-dimensional image data representing a two-dimensional image; the eigenvalue calculation step is performed by one or more kinds of algorithms, and the feature values of the pixels constituting the two-dimensional image data are based on the brightness values of the pixels Calculating a processed image data generating step of dividing each pixel constituting the two-dimensional image data into a defective pixel having the feature value equal to or greater than a predetermined threshold value, and remaining pixels having the feature value not reaching the threshold value, and generating Processing the image data, the processed image data includes a grayscale information storage bit string storing grayscale information indicating a grayscale value corresponding to the feature value for the defective pixel, and storing the remaining pixel for the remaining pixel a gray scale information storage bit string representing gray scale information of a gray scale value of zero; a defect information acquisition step of acquiring defect information about a defect in the sheet shaped body for each pixel according to the processed image data described above, and Generating a defect information storage bit string storing the defect information acquired as described above; and analyzing the image data generating step, which generates an image for analysis for each pixel Data, the parsing image data includes a parsing bit string obtained by adding the defect information storage bit string to the grayscale information storage bit string of the processed image data; and an image parsing step, which is stored and used The information of the analysis bit string constituting the analysis image data is subjected to predetermined image analysis, thereby detecting defects of the sheet-shaped molded body.

According to the present invention, an image generating apparatus generates an apparatus for inspecting image data of a defect of a sheet-shaped formed body, and includes a conveying unit, a light irradiation unit, an imaging unit, a feature value calculation unit, and a processed image data generation unit, The defect information acquisition unit and the analysis image data generation unit. In the image generating apparatus, the imaging unit performs an imaging operation on the sheet-shaped molded body conveyed by the transport unit while irradiating light by the light-irradiating portion, and generates two-dimensional image data representing the two-dimensional image. The feature value calculation unit processes the two-dimensional image data by a predetermined algorithm, thereby calculating a feature value based on the luminance value of each pixel constituting the two-dimensional image data.

The processed image data generating unit divides each pixel constituting the two-dimensional image data into a defective pixel whose feature value is equal to or greater than a predetermined threshold value, and a remaining pixel whose feature value does not reach the threshold value, and generates a processed image. And the processed image data includes a grayscale information storage bit string storing grayscale information indicating a grayscale value corresponding to the feature value for the defective pixel, and storing the zero indicating for the remaining pixels Gray scale information storage bit string of gray scale information of gray scale value. The defect information acquisition unit acquires information on defects in the sheet-shaped formed body, that is, defect information, for each pixel, based on the processed image data, and generates a defect information storage bit string storing the acquired defect information.

The analysis image data generation unit generates analysis image data for each pixel, and the analysis image data includes a defect information acquisition unit that is added to the grayscale information storage bit string of the processed image data. The above-mentioned defect information is stored in a bit string manner to obtain a parsing bit string.

In the image generating apparatus of the present invention configured as described above, the image data for inspecting the defect of the sheet-shaped formed body, that is, the analysis map, is generated based on the two-dimensional image data of the sheet-shaped molded body produced by the image forming unit. Like the data, it is possible to maintain a high defect detection capability as compared with a case where image data for checking defects is generated based on, for example, a plurality of one-dimensional image data obtained by using a line sensor.

Further, in the image generating apparatus of the present invention, the two-dimensional image data having a large amount of information output from the image capturing unit is converted into processed image data constituting each pixel by using the grayscale information storage bit string, and further converted into utilization. The parsing bit string obtained by adding the defect information storage bit string to the gray scale information storage bit string constitutes parsing image data for each pixel. The image generating apparatus generates image data for analysis which constitutes each pixel by the analysis bit string converted from the two-dimensional image data in this manner, and is used as image data for inspecting defects of the sheet-shaped formed body. By performing image analysis using the analysis image data, it is possible to speed up image analysis and improve the efficiency of defect inspection.

Moreover, according to the present invention, the defect information acquired by the defect information acquiring unit based on the processed image data may include defect type information indicating the type of the defect in the sheet-shaped formed body. Thereby, the image generating apparatus can acquire information related to the kind of the defect in the sheet-shaped formed body based on the defect type information.

Further, according to the present invention, the feature value calculation unit calculates the feature value by processing by a plurality of algorithms. Moreover, the defect information acquisition unit may store, according to the grayscale information of each pixel, whether the grayscale information of the bit string is grayscale information corresponding to the feature value calculated by processing any one of the plurality of algorithm processes. And obtain defect information containing information on the type of defect.

Moreover, according to the present invention, the defect inspection device includes the image generation device and the image analysis device of the present invention described above. The image analysis device performs predetermined image analysis using the information of the analysis bit string stored in the analysis image data generated by the analysis image data generating unit of the image generation device. The defects of the sheet-like formed body were examined. As a result, it is possible to increase the speed of image analysis by the image analysis device, and it is possible to improve the efficiency of defect inspection.

Moreover, according to the present invention, the defect inspection method includes a transfer step, a light irradiation step, a photographing step, a feature value calculation step, a processed image data generation step, a defect information acquisition step, an analysis image data generation step, and an image analysis step. . In the defect inspection method, in the imaging step, the sheet-shaped molded body conveyed by the transport unit while being irradiated with light is subjected to an image capturing operation, and two-dimensional image data representing the two-dimensional image is generated. In the feature value calculation step, the two-dimensional image data is processed by a predetermined algorithm to calculate a feature value based on the luminance value of each pixel constituting the two-dimensional image data.

In the processing image data generating step, each pixel constituting the two-dimensional image data is divided into a defective pixel whose feature value is equal to or greater than a predetermined threshold value, and a remaining pixel whose feature value does not reach the threshold value, and is generated and processed. Image data, the processed image data includes a grayscale resource stored for the defective pixel and representing a grayscale value corresponding to the feature value The grayscale information storage bit string of the message includes a grayscale information storage bit string storing grayscale information indicating the grayscale value of zero for the remaining pixels. In the defect information obtaining step, the defect information about the defect in the sheet-shaped formed body is acquired for each pixel according to the processed image data, and the defect information storage bit string storing the obtained defect information is generated.

In the image forming step of parsing, image data for parsing is generated for each pixel, and the image data for parsing includes adding the defect information storage bit to the grayscale information storage bit string of the processed image data. The parsing bit string obtained by the string method. Then, in the image analysis step, the predetermined image analysis is performed using the information stored in the analysis bit string constituting the analysis image data, thereby detecting the defect of the sheet-shaped molded body.

In the defect inspection method of the present invention configured as described above, the defect detection of the sheet-shaped formed body is performed based on the two-dimensional image data of the sheet-like formed body generated in the image capturing step, and thus, for example, according to the use of the line sensor Compared with the case where a plurality of one-dimensional image data are obtained for defect detection, a high defect detection capability can be maintained.

Further, in the defect inspection method of the present invention, the two-dimensional image data having a large amount of information generated in the photographing step is converted into processed image data constituting each pixel by using the gray scale information storage bit string, and further converted into The parsing bit string obtained by adding the defect information storage bit string to the gray scale information storage bit string constitutes parsing image data for each pixel. According to the analysis image data for each pixel formed by the analysis bit string converted from the two-dimensional image data in this way, image analysis is performed in the image analysis step to detect the defect of the sheet-shaped molded body, and thus the defect is detected. The image analysis speed in the image analysis step can be increased, and the inspection efficiency can be improved.

1‧‧‧Image generating device

2‧‧‧Sheet shaped body

3‧‧‧Transporting device

4‧‧‧Lighting device

5‧‧‧Photographing device

6‧‧‧Image processing device

7‧‧‧Image analysis device

61‧‧‧Processing Image Generation Department

62‧‧‧Image generation unit for analysis

71‧‧‧Image input unit for analysis

72‧‧‧Image Analysis Department

73‧‧‧Control Department

74‧‧‧Display Department

100‧‧‧ Defect inspection device

A‧‧‧data point

A‧‧‧ two-dimensional image

A1‧‧‧ illumination image

A3‧‧‧ upper limit edge

A4‧‧‧ lower edge

A21‧‧‧1st defective pixel group

A22‧‧‧2nd defective pixel group

B‧‧‧data points

B‧‧‧2D image

B1‧‧‧Lighting images

B21‧‧‧1st defective pixel group

B22‧‧‧2nd defective pixel group

C‧‧‧data points

C‧‧‧2D image

C1‧‧‧ illumination image

C21‧‧‧1st defective pixel group

C22‧‧‧2nd defective pixel group

C31‧‧‧ core

D‧‧‧data points

D‧‧‧ two-dimensional image

D1‧‧‧ illumination image

D3‧‧‧ edge

D21‧‧‧1st defective pixel group

D22‧‧‧2nd defective pixel group

E‧‧‧Processing images

E21‧‧‧1st defective pixel group

E22‧‧‧ Remaining pixel group

F‧‧‧ forecast data points

F‧‧‧Processing images

F21‧‧‧2nd defective pixel group

F22‧‧‧ Remaining pixel group

G‧‧‧Processing images

G21‧‧‧1st defective pixel group

G22‧‧‧2nd defective pixel group

G23‧‧‧ Remaining pixel group

G31‧‧‧ Grayscale information storage bit string

G32‧‧‧ Grayscale information storage bit string

G33‧‧‧ Grayscale information storage bit string

H‧‧‧Image for analysis

H21‧‧‧1st defective pixel group

H22‧‧‧2nd defective pixel group

H23‧‧‧Remaining pixel group

H31‧‧‧ parsing bit string

H32‧‧‧ parsing bit string

H33‧‧‧ parsing bit string

H311‧‧‧ Grayscale information storage bit string

H312‧‧‧ Defect information storage bit string

H321‧‧‧ Grayscale information storage bit string

H322‧‧‧ Defect information storage bit string

H331‧‧‧ Grayscale information storage bit string

H332‧‧‧ Defect information storage bit string

L‧‧‧ Straight line

P1‧‧‧ edge distribution

P2‧‧‧ differential distribution

P3‧‧‧Brightness distribution

P4‧‧‧Brightness value difference distribution

P5‧‧‧ Smoothing distribution

P6‧‧‧ edge distribution

P7‧‧‧ edge distribution

P11‧‧‧ peak

P21‧‧‧ peak

P22‧‧‧ eigenvalue

P31‧‧‧ Valley section

P41‧‧‧ peak

P42‧‧‧ eigenvalue

P51‧‧‧ peak

P52‧‧‧ eigenvalue

P61‧‧ points

P62‧‧‧ points

P63‧‧‧ area

P71‧‧ points

P72‧‧ points

P73‧‧‧ imaginary straight line

P74‧‧‧Arc

P711‧‧‧ tangent

P721‧‧‧ tangent

R‧‧‧ radius of curvature

X‧‧‧ direction

Y‧‧‧ direction

Z‧‧‧Transfer direction

11‧‧‧ angle

22‧‧‧ angle

33‧‧‧ angle

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a flow chart showing the steps of a defect inspection method according to an embodiment of the present invention.

Fig. 2 is a schematic view showing the configuration of a defect inspection apparatus 100 according to an embodiment of the present invention.

FIG. 3 is a block diagram showing the configuration of the defect inspection apparatus 100.

4A is a view for explaining an edge distribution method as an example of a defect detection algorithm, and is a view showing an example of a two-dimensional image A corresponding to two-dimensional image data generated by the imaging device 5.

FIG. 4B is a view showing an example of the edge distribution P1 formed by the processed image generating unit 61.

4C is a view showing an example of the differential distribution P2 made by the processed image generating unit 61.

5A is a view for explaining a peak method as another example of the defect detection algorithm, and is a view showing an example of a two-dimensional image B corresponding to the two-dimensional image data generated by the imaging device 5.

FIG. 5B is a view showing an example of the luminance distribution P3 created by the processed image generating unit 61.

Fig. 5C is a diagram for explaining a hypothetical sequence of the mass points which are executed by the processed image generating unit 61 from one end of the data point toward the other end.

FIG. 5D is a view showing an example of the luminance value difference distribution P4 made by the processed image generating unit 61.

6A is a view for explaining a smoothing method as another example of the defect detecting algorithm, and is a view showing an example of a two-dimensional image C corresponding to the two-dimensional image data generated by the imaging device 5. .

FIG. 6B is a view showing an example of the smoothing distribution P5 generated by the processed image generating unit 61.

7A is a diagram for explaining a second edge distribution method as another example of the defect detection algorithm, and corresponds to a two-dimensional image data generated by the imaging device 5. A diagram of an example of a two-dimensional image D.

Fig. 7B is a view showing an example of the edge distribution P6 formed by the processed image generating unit 61.

Fig. 7C is a view showing an example of the edge distribution P7 formed by the processed image generating unit 61.

FIG. 8A is a view showing an example of a processed image generated by the image processing device 6, and is a view showing an example of the processed image E generated by the first defect detecting algorithm.

8B is a view showing an example of a processed image generated by the image processing device 6, and is a view showing an example of the processed image F generated by the second defect detecting algorithm.

FIG. 8C is a view showing an example of the processed image G generated by the processed image generating unit 61 combining the processed image E and the processed image F.

FIG. 9A is a view showing an example of an image for analysis generated by the image processing device 6, and shows grayscale information storage by each pixel constituting the processed image G generated by the processed image generating portion 61. A diagram of an example of the analysis image H obtained by adding a defect information to the bit string.

FIG. 9B is a view showing an example of the analysis bit strings H31, H32, and H33 constituting the pixels in the analysis image H.

BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a flow chart showing the steps of a defect inspection method according to an embodiment of the present invention. The defect inspection method of the present embodiment includes the transport step s1, the light irradiation step s2, the photographing step s3, the feature value calculating step s4, the processed image data generating step s5, the defect information acquiring step s6, and the parsing image shown in Fig. 1. The data generation step s7 and the image analysis step s8.

Figure 2 is a view showing the configuration of the defect inspection apparatus 100 according to an embodiment of the present invention. Figure. FIG. 3 is a block diagram showing the configuration of the defect inspection apparatus 100. The defect inspection device 100 of the present embodiment is a device for detecting a defect of the sheet-like molded body 2 such as a thermoplastic resin, and includes the image generating device 1 and the image analyzing device 7 of the present invention. The image generation device 1 of the defect inspection device 100 includes a transfer device 3, an illumination device 4, an imaging device 5, and an image processing device 6. The defect inspection device 100 implements the defect inspection method of the present invention. The transport device 3 performs the transport step s1, the illumination device 4 performs the light irradiation step s2, the photographing device 5 performs the photographing step s3, and the image processing device 6 performs the feature value calculation step s4, the processed image data generation step s5, and the defect information acquisition step s6. And the analysis image data generation step s7, the image analysis device 7 performs an image analysis step s8.

In the defect inspection apparatus 100, the sheet-like molded body 2 continuous in the longitudinal direction with a constant width is transferred in the fixed direction (the same direction as the longitudinal direction orthogonal to the width direction of the sheet-shaped molded body 2) by the conveyance device 3, And the two-dimensional image data representing the two-dimensional image is generated by the photographing device 5 capturing the sheet surface illuminated by the illumination device 4 during the transfer, and the image processing device 6 generates the image according to the two-dimensional image data. The image data for analysis is used, and the image analysis device 7 performs defect detection based on the analysis image data output from the image processing device 6.

The sheet-like molded body 2 as the object to be inspected is formed by subjecting a thermoplastic resin extruded from an extruder to a surface of a roll to smooth the surface or impart a treatment such as a concavo-convex shape, and to carry it on the conveying roller. The upper cooling side is taken up by a take-up roll. The thermoplastic resin which can be applied to the sheet-like formed body 2 of the present embodiment is, for example, a methacrylic resin, a methyl methacrylate styrene copolymer (MS), or a polyethylene (PE). Polypropylene (PP, polypropylene) and other polyolefins, polycarbonate (PC), polyvinyl chloride (PVC), polystyrene (PS, polystyrene), polyvinyl alcohol (PVA, polyvinyl alcohol), Triacetyl cellulose resin (TAC, Triacetyl Cellulose) and the like. The sheet-shaped formed body 2 is a single-layer sheet or laminated layer using the thermoplastic resins Formed by sheet or the like.

Further, examples of the defects generated in the sheet-like molded body 2 include dot-shaped defects (point defects) such as bubbles, fish eyes, foreign matter, tire marks, dents, and scratches generated during molding, and creases and the like. The so-called knick, a linear defect such as a so-called original stripe which is caused by a difference in thickness (line defect).

The conveying device 3 has a function as a conveying unit, and conveys the sheet-shaped formed body 2 in the fixing direction (transporting direction Z). The conveying device 3 includes, for example, a feeding roller and a receiving roller that convey the sheet-shaped formed body 2 in the conveying direction Z, and measures the conveying distance by a rotary encoder or the like. In the present embodiment, the conveying speed is set to about 2 to 30 m/min in the conveying direction Z.

The illuminating device 4 has a function as a light-irradiating portion, and has a line-like illumination in the width direction of the sheet-like molded body 2 orthogonal to the transport direction Z. The illuminating device 4 is disposed such that the image captured by the imaging device 5 includes a linear reflected image. Specifically, the illuminating device 4 is formed on the same side as the imaging device 5 with respect to the sheet-like molded body 2, and faces the surface of the sheet-like molded body 2 and is separated from the illumination region on the surface of the sheet-like molded body 2, that is, photographed. The distance between the imaging regions captured by the device 5 is set to, for example, 200 mm.

As a light source of the illumination device 4, light that does not affect the composition and properties of the sheet-like formed body 2 is irradiated with an LED (Light Emitting Diode), a metal halide lamp, a halogen transmission lamp, a fluorescent lamp, or the like. However, there is no particular limitation. Further, the illumination device 4 may be disposed on the opposite side of the imaging device 5 via the sheet-like molded body 2. In this case, the image captured by the imaging device 5 includes the transmitted image transmitted through the sheet-like formed body 2.

The defect inspection apparatus 100 includes a plurality of imaging devices 5 having a function as an imaging unit, and each imaging device 5 is arranged at equal intervals in a direction orthogonal to the transport direction Z (the width direction of the sheet-shaped molded body 2). Further, the imaging device 5 is disposed such that the direction from the imaging device 5 toward the center of the imaging region of the sheet-like molded body 2 forms an acute angle with the transport direction Z. The imaging device 5 generates a plurality of two-dimensional images including the reflected image or the transmitted image (hereinafter collectively referred to as "illuminated image") generated by the illumination device 4 in a plurality of times. Dimensional image data.

The imaging device 5 includes a CCD (Charge Coupled Device) or a CMOS (Complementary Metal-Oxide Semiconductor) area sensor that captures a two-dimensional image. As shown in FIG. 2, the imaging device 5 is disposed so as to capture the entire region of the sheet-like molded body 2 in the width direction orthogonal to the transport direction Z. By photographing the entire region in the width direction of the sheet-like molded body 2 in this manner and transporting the sheet-like molded body 2 continuous in the transport direction Z, the defects of the entire region of the sheet-like molded body 2 can be efficiently inspected.

The photographing interval (frame rate) of the photographing device 5 may be fixed or may be changed by the user operating the photographing device 5 itself. Further, the photographing interval of the photographing device 5 may be a fraction of a second as a time interval of continuous photographing of the digital still camera, but in order to improve the efficiency of the inspection, it is preferably a short time interval, for example, as a general The frame rate of the animation data is 1/30 second, etc.

The length of the transport direction Z of the two-dimensional image captured by the imaging device 5 is preferably such that the sheet-shaped formed body 2 is transported after the photographing device 5 takes in the two-dimensional image and before the next two-dimensional image is taken in. The transport distance is at least 2 times or more. That is, it is preferable to photograph the same portion of the sheet-like formed body 2 twice or more. In this manner, the length of the transport direction Z of the two-dimensional image is larger than the transport distance of the sheet-like molded body 2 in the time before the image pickup device 5 takes in the two-dimensional image and takes the next two-dimensional image. The number of times of photographing of the same portion of the sheet-like formed body 2 is increased, whereby the defect can be inspected with high precision.

The image processing device 6 includes a processed image generating unit 61 having functions as a feature value calculating unit, a processed image data generating unit, and a defect information acquiring unit, and an analysis image generating unit 62 having analysis information. The function of the image data generation section. The image processing device 6 is provided corresponding to each of the plurality of imaging devices 5. The processed image generating unit 61 can be implemented by an FPGA (Field-programmable gate array) or a GPGPU (General-purpose computing on graphics processing units). The graphics processing unit or the like is implemented by an image processing board or a hardware inside the imaging device 5.

The processed image generating unit 61 processes the two-dimensional image data output from the imaging device 5 by using a predetermined algorithm (hereinafter referred to as "defect detection algorithm"), thereby calculating the two-dimensional image. The characteristic value of each pixel of the data based on the luminance value. Further, the processed image generating unit 61 identifies the pixel having the feature value equal to or greater than a predetermined threshold value as a defective pixel in the two-dimensional image data, and generates a grayscale information storage bit string, the grayscale information storage bit In the string, gray scale information indicating a gray scale value corresponding to the feature value is stored for the defective pixel, and a gray scale indicating zero is stored for the remaining pixels other than the defective pixel (the pixel whose feature value does not reach the threshold value) Grayscale information of the value. The gray scale information storage bit string generated for each pixel respectively contains a plurality of bits. Then, the processed image generating unit 61 outputs processed image data each of which is composed of the above-described grayscale information storage bit string. Further, the processed image generating unit 61 acquires information on the defect in the sheet-shaped formed body 2, that is, defect information, for each pixel, based on the processed image data generated, and generates defect information storing the acquired defect information. Store the bit string. The defect information storage bit string generated for each pixel typically contains a plurality of bits, respectively.

The defect detection algorithm used in the processed image generating unit 61 will be described with reference to FIGS. 4A to 4C, FIGS. 5A to 5D, FIGS. 6A and 6B, and FIGS. 7A to 7C.

4A to 4C are diagrams for explaining an edge distribution method as an example of a defect detection algorithm. 4A shows an example of a two-dimensional image A corresponding to the two-dimensional image data generated by the imaging device 5, and the upper side of the image is the downstream side of the transport direction Z, and the lower side of the image is the upstream side of the transport direction Z. In the two-dimensional image A, the direction parallel to the width direction of the sheet-like molded body 2 is set to the X direction, and the direction parallel to the longitudinal direction of the sheet-shaped molded body 2 (the direction parallel to the conveyance direction Z) is set to Y direction. In FIG. 4A, the band-shaped bright region which is located at the center in the Y direction of the two-dimensional image A and extends in the X direction is the illumination image A1, and the dark region existing inside the illumination image A1 is the first defective pixel group A21. The bright region existing in the vicinity of the illumination image A1 is the second defective pixel group A22.

When the defect detection algorithm using the edge distribution method is used, the processed image generating portion 61 first divides the two-dimensional image A into data of progressive pixel rows along the Y direction. Next, the processed image generating portion 61 traces the data of each pixel row from one end in the Y direction (the upper end of the two-dimensional image A in Fig. 4A) toward the other end (the lower end of the two-dimensional image A in Fig. 4A). Edge determination processing.

Specifically, the processed image generating unit 61 sets the second pixel from the one end side in the Y direction as the target pixel for each pixel row, and determines whether the brightness value of the target pixel is on the one end side with respect to the target pixel. The luminance values of adjacent pixels adjacent to each other are greater than a certain threshold. When it is determined that the luminance value of the pixel of interest is larger than the luminance value of the adjacent pixel by a certain threshold or more, the processed image generation unit 61 determines that the adjacent pixel is the upper limit edge A3. In other cases, the processed image generating unit 61 repeats the edge determination process while shifting the pixel of interest toward the other end in the Y direction by one pixel until the brightness value of the pixel of interest is determined to be higher than the brightness value of the adjacent pixel. The maximum specific threshold is above.

After detecting the upper limit edge A3, the processed image generating unit 61 determines whether the luminance value of the pixel of interest is smaller than the luminance value of the adjacent pixel by a specific threshold or more by shifting the pixel of interest toward the other end in the Y direction by one pixel. . When it is determined that the luminance value of the pixel of interest is smaller than the threshold value of the adjacent pixel by a specific threshold or more, the processed image generation unit 61 determines that the adjacent pixel is the lower limit edge A4. In other cases, the processed image generating unit 61 repeats the edge determination process while shifting the pixel of interest toward the other end in the Y direction by one pixel until the brightness value of the pixel of interest is determined to be higher than the brightness value of the adjacent pixel. Small specific thresholds or more.

In the example of FIG. 4A, the upper limit edge A3 detected by the edge determination processing of the processed image generating unit 61 is shown by "○", and the lower limit edge A4 is shown by "●". 4A, in the two-dimensional image A, the coordinate value (Y coordinate value) of the upper limit edge A3 and the lower limit edge A4 in the Y direction in the first defective pixel group A21 and the second defective pixel group A22 in which the defect exists The difference is extremely smaller than the Y coordinate in the remaining pixels except the defective pixel The difference between the values. Further, in the second defective pixel group A22 in the two-dimensional image A, the Y coordinate value of the upper limit edge A3 is significantly different from the Y coordinate value of the remaining pixels other than the defective pixel.

With such a feature, the processed image generating portion 61 creates the edge distribution P1 shown in Fig. 4B. In the edge distribution P1 shown in FIG. 4B, a peak P11 corresponding to the Y coordinate value of the upper limit edge A3 appears corresponding to the second defective pixel group A22 in the two-dimensional image A. Furthermore, the processed image generating unit 61 may be configured to form an edge distribution based on the difference between the Y coordinate values of the upper limit edge A3 and the lower limit edge A4. In this case, in the edge distribution by the processed image generating unit 61, the upper limit edge A3 and the lower limit appear corresponding to the first defective pixel group A21 and the second defective pixel group A22 in the two-dimensional image A. The difference between the Y coordinate values of the edge A4 is small.

Further, the processed image generating unit 61 differentiates the edge distribution P1 to form a differential distribution P2 shown in FIG. 4C. In the differential distribution P2 shown in FIG. 4C, corresponding to the peak value P11 in the edge distribution P1, that is, corresponding to the second defective pixel group A22 in the two-dimensional image A, a predetermined threshold or more appears (differential The value of the larger value is the peak value P21 of the feature value P22.

The processed image generating unit 61 extracts a pixel in the two-dimensional image A corresponding to the peak value P21 of the feature value P22 having a predetermined threshold value or more as a defective pixel based on the differential distribution P2. In the example of the differential distribution P2 shown in FIG. 4C, the processed image generating unit 61 extracts the second defective pixel group A22 as a defective pixel.

5A to 5D are diagrams for explaining a peak method as another example of the defect detection algorithm. FIG. 5A shows an example of a two-dimensional image B corresponding to the two-dimensional image data generated by the imaging device 5, and the upper side of the image is the downstream side of the transport direction Z, and the lower side of the image is the upstream side of the transport direction Z. In the two-dimensional image B, the direction parallel to the width direction of the sheet-like molded body 2 is set to the X direction, and the direction parallel to the longitudinal direction of the sheet-shaped molded body 2 (the direction parallel to the conveyance direction Z) is set to Y direction. In FIG. 5( a ), the band-shaped bright region located at the center in the Y direction of the two-dimensional image B and extending in the X direction is the illumination image B1 and exists in the photo. The dark area inside the bright image B1 is the first defective pixel group B21, and the bright area existing in the vicinity of the illumination image B1 is the second defective pixel group B22.

When the defect detection algorithm using the peak method is used, the processed image generating portion 61 first divides the two-dimensional image B into data of progressive pixel rows along the Y direction. Next, the processed image generating unit 61 is configured to continuously draw and connect the data of the luminance values at the position on the straight line L parallel to the Y direction of the two-dimensional image B for each pixel row. The curve formed by this is the luminance distribution P3 shown in FIG. 5B.

In the case where there is no defective pixel in the two-dimensional image B, the luminance distribution P3 exhibits a distribution of a single peak where no valley portion occurs, and in the case where the defective pixel exists, as shown in FIG. 5B, the double of the valley portion P31 appears. The distribution of peaks.

Then, the processed image generating unit 61 assumes that the moving time between adjacent data points is fixed regardless of the distance between the data points, regardless of the distance between the data points, from one end of the X direction of the brightness distribution P3 toward the other end. The point of movement. Here, the mass point is moved from the data point c toward the data point b adjacent thereto from the data point c, toward the data point a adjacent thereto from the data point b, and from the data point a toward the data point d adjacent thereto. Further, the data point d is set as the data point corresponding to the pixel of interest.

The processed image generating unit 61 obtains the velocity vector and the acceleration vector of the mass point at the data points a, b, and c passing through the material point d immediately before passing through the data point d. In other words, the processed image generating unit 61 obtains the dot in the interval from the data point b to the data point a based on the coordinates of the two data points a and the data point b that are passed before the data point d and the moving time. Speed vector. Further, the processed image generating unit 61 obtains the mass point in the interval from the data point c to the data point b based on the coordinates of the two data points b and the data points c which are passed before the data point a and the moving time. Speed vector. Further, the processed image generating unit 61 obtains the data point c to the data point based on the velocity vector of the mass point in the interval from the data point b to the data point a and the velocity vector of the mass point in the interval from the data point c to the data point b. The acceleration vector of the particle in the interval of a. Then, the processed image generating unit 61 is within the interval from the data point b to the data point a The velocity vector of the mass point and the acceleration vector of the mass point in the interval from the data point c to the data point a predict the coordinates of the data point d (predicted data point f).

The processed image generating unit 61 obtains the difference between the luminance value of the predicted data point f of the data point d predicted in the above manner and the actual (actually measured) luminance value of the data point d, and creates the luminance value difference shown in FIG. 5D. Distribution P4. In the luminance value difference distribution P4 shown in FIG. 5D, corresponding to the valley portion P31 in the luminance distribution P3 shown in FIG. 5B, that is, corresponding to the first defective pixel group B21 in the two-dimensional image B, appears The peak value P41 of the characteristic value P42 having a predetermined threshold value or more (the luminance value difference is large).

The processed image generating unit 61 extracts a pixel in the two-dimensional image B corresponding to the peak value P41 of the feature value P42 having a predetermined threshold value or more as a defective pixel based on the luminance value difference distribution P4. In the example of the luminance value difference distribution P4 shown in FIG. 5D, the processed image generating unit 61 extracts the first defective pixel group B21 as a defective pixel.

6A and 6B are diagrams for explaining a smoothing method as another example of the defect detection algorithm. FIG. 6A shows an example of a two-dimensional image C corresponding to the two-dimensional image data generated by the imaging device 5, and the upper side of the image is on the downstream side in the transport direction Z, and the lower side of the image is on the upstream side in the transport direction Z. In the two-dimensional image C, the direction parallel to the width direction of the sheet-like molded body 2 is set to the X direction, and the direction parallel to the longitudinal direction of the sheet-shaped molded body 2 (the direction parallel to the conveyance direction Z) is set to Y direction. In FIG. 6A, the strip-shaped bright region located in the center of the two-dimensional image C in the Y direction and extending in the X direction is the illumination image C1, and the dark region existing inside the illumination image C1 is the first defective pixel group C21. The bright region existing in the vicinity of the illumination image C1 is the second defective pixel group C22.

When the defect detection algorithm using the smoothing method is used, the processed image generating portion 61 first divides the two-dimensional image C into data of progressive pixel rows along the Y direction. Next, the processed image generating unit 61 is formed as a core C31 having a plurality of pixels (for example, five pixels in the X direction and one pixel in the Y direction) in the X direction and the Y direction.

Then, the processed image generating unit 61 is configured to create a two-dimensional image C for each pixel row. The data of the difference between the luminance value of the central pixel in the core C31 at the position on the straight line L parallel to the Y direction and the average value of the luminance values of all the pixels in the core C31 is continuously drawn and connected in the form of dots. The curve formed by this is the smoothing distribution P5 shown in Fig. 6B. In the smoothing distribution P5 shown in FIG. 6B, a peak P51 having a feature value P52 having a predetermined threshold or more (a large difference in luminance value) appears in correspondence with the first defective pixel group C21 in the two-dimensional image C. .

The processed image generating unit 61 extracts, as a defective pixel, a pixel in the two-dimensional image C corresponding to the peak value P51 of the feature value P52 having a predetermined threshold value or more, based on the smoothing distribution P5. In the example of the smoothing distribution P5 shown in FIG. 6B, the processed image generating unit 61 extracts the first defective pixel group C21 as a defective pixel.

7A to 7C are diagrams for explaining a second edge distribution method as another example of the defect detection algorithm. FIG. 7A shows an example of a two-dimensional image D corresponding to the two-dimensional image data generated by the imaging device 5, and the upper side of the image is the downstream side of the transport direction Z, and the lower side of the image is the upstream side of the transport direction Z. In the two-dimensional image D, the direction parallel to the width direction of the sheet-like molded body 2 is set to the X direction, and the direction parallel to the longitudinal direction of the sheet-shaped molded body 2 (the direction parallel to the conveyance direction Z) is set to Y direction. In FIG. 7A, a bright region in the center of the two-dimensional image D in the Y direction and extending in the X direction is an illumination image D1, and a dark region existing inside the illumination image D1 is a first defective pixel group D21. The bright region existing in the vicinity of the illumination image D1 is the second defective pixel group D22.

When the defect detection algorithm using the second edge distribution method is used, the processed image generation unit 61 first divides the two-dimensional image D into data of progressive pixel rows along the Y direction. Next, the processed image generating unit 61 detects the data of each pixel row from one end in the Y direction (the upper end of the two-dimensional image D in Fig. 7A) toward the other end (the lower end of the two-dimensional image D in Fig. 7A). Edge edge determination processing.

Specifically, the processed image generating unit 61 sets the second pixel from the one end side in the Y direction as the target pixel for each pixel row, and determines whether or not the luminance value of the pixel of interest is The luminance value of the adjacent pixel adjacent to the one end side with respect to the pixel of interest is larger than a threshold value. When it is determined that the luminance value of the pixel of interest is larger than the luminance value of the adjacent pixel by a certain threshold or more, the processed image generation unit 61 determines that the adjacent pixel is the edge D3. In other cases, the processed image generating unit 61 repeats the edge determination process while shifting the pixel of interest toward the other end in the Y direction by one pixel until the brightness value of the pixel of interest is determined to be higher than the brightness value of the adjacent pixel. Above a certain threshold.

In FIG. 7A, an example in which the edge D3 detected by the edge determination processing by the processed image generating unit 61 is detected is indicated by "○". 7A, in the boundary portion between the bright region and the dark region of the two-dimensional image D, in the second defective pixel group D22 in which the defect exists, the coordinate value (Y coordinate value) of the edge D3 in the Y direction is extremely extreme. Variety.

There are two methods for extracting defective pixels in the two-dimensional image D by using such a feature. In the first method shown in FIG. 7B, the processed image generating portion 61 creates an edge distribution P6 corresponding to the edge D3 in the two-dimensional image D. Further, in FIG. 7B, the edge distribution P6 corresponding to the edge D3 in the vicinity of the second defective pixel group D22 of the two-dimensional image D is enlarged. In the edge distribution P6 shown in FIG. 7B, the Y coordinate value extremely changes in correspondence with the second defective pixel group D22 in the two-dimensional image D.

The processed image generating unit 61 selects a point P61 and a point P62 which are arbitrary two points on the edge distribution P6 that has been created, and calculates a region P63 surrounded by a curve connecting the line connecting the point P61 and the point P62 with the edge distribution P6. The area is taken as the eigenvalue. The processed image generating unit 61 extracts, as a defective pixel, a pixel in the two-dimensional image D corresponding to the distribution portion having the feature value (area of the region P63) having a predetermined threshold value or more based on the edge distribution P6.

In the second method shown in FIG. 7C, the processed image generating unit 61 creates an edge distribution P7 corresponding to the edge D3 of the two-dimensional image D. In addition, in FIG. 7C, the edge distribution P7 corresponding to the edge D3 in the vicinity of the second defective pixel group D22 of the two-dimensional image D is enlarged. In the edge distribution P7 shown in FIG. 7C, the Y coordinate value extremely changes in correspondence with the second defective pixel group D22 in the two-dimensional image D.

The processed image generating unit 61 selects the point P71 and the point P72 which are any two points on the edge distribution P7 which is formed, and creates a tangent to the tangent P711 of the edge distribution P7 at the point P71 and the edge distribution P7 at the point P72. P721. Next, the processed image generating unit 61 calculates an angle α1 between the virtual straight line P73 parallel to the X-axis and the tangent line P711, and an angle α2 between the virtual straight line P73 and the tangent line P721, and obtains the calculated angle α1 as the above. The angle α3 of the difference of the angle α2. Then, the processed image generating portion 61 calculates the radius of curvature of the arc P74 between the point P71 and the point P72 in the edge distribution P7 using the length of the arc P74 between the point P71 and the point P72 in the edge distribution P7 and the angle α3. R is used as the eigenvalue. The processed image generating unit 61 extracts pixels in the two-dimensional image D corresponding to the distribution portion having the feature value (curvature radius R) within the predetermined threshold range as the defective pixel based on the edge distribution P7.

As the defects generated in the sheet-like molded body 2, as described above, point defects such as bubbles, fish eyes, foreign matter, tire marks, dents, and scratches, and so-called knicks due to creases and the like are exemplified. A line defect such as a so-called original stripe which is caused by a difference in thickness.

The type of the defect that can be extracted differs depending on the type of the defect detection algorithm used when the processed image generation unit 61 generates the processed image. As the edge distribution method which is an example of the defect detection algorithm, it is possible to extract defects such as foreign matter, tire marks, scratches, and the like with a high extraction ability. The above peak method can extract defects such as foreign matter, dents, scratches and the like with a high extraction ability. The above smoothing method can extract defects such as bubbles, fish eyes, and dents with a high extraction ability. The second edge distribution method described above can extract defects such as stripe or cracks of the original sheet with a high extraction capability.

The processed image generating unit 61 calculates the feature value by using the processing of the plurality of defect detecting algorithms by using the difference in the defect extracting ability caused by the type of the defect detecting algorithm as described above. Then, the defective pixel in the two-dimensional image is extracted using the above-described calculated feature value, whereby the defect type of the defective region in the two-dimensional image generated by the imaging device 5 can be distinguished.

8A to 8C are views showing an example of a processed image generated by the image processing device 6. In the present embodiment, the processed image generating unit 61 of the image processing device 6 processes the two-dimensional image data output from the imaging device 5 by the defect detecting algorithm, and extracts defective pixels, and then generates a defect as shown in FIG. 8A. The processed image shown in ~8C. In the example shown in FIGS. 8A to 8C, the processed image generating unit 61 extracts a two-dimensional image using the first defect detecting algorithm and the second defect detecting algorithm which are two kinds of defect detecting algorithms of different types. The defective pixel produces a processed image. Here, the first defect detection algorithm has a high extraction ability for the first defective pixel group in the two-dimensional image generated by the imaging device 5, but does not have the extraction capability for the second defective pixel group. Further, the second defect detection algorithm has a high extraction ability for the second defective pixel group in the two-dimensional image generated by the imaging device 5, but does not have the extraction capability for the first defective pixel group.

The processed image generating unit 61 processes the two-dimensional image data output from the imaging device 5 in parallel by the first defect detecting algorithm and the second defect detecting algorithm, and extracts the feature value to a predetermined threshold or more (for use). When the second edge distribution method is used as the defect detection algorithm, the pixel of the "feature value is within a predetermined threshold range" is used as the defective pixel, and the processed image E as shown in FIG. 8A is generated, and as shown in FIG. 8B. The image F is processed as shown.

The processed image E shown in FIG. 8A is a processed image generated by processing by the first defect detecting algorithm, and is composed of the following grayscale information storage bit string, that is, the first defect detecting algorithm can be used The first defective pixel group E21 extracted by the processing stores grayscale information indicating a grayscale value corresponding to the feature value, and stores a grayscale value indicating zero for the remaining pixel group E22 other than the first defective pixel group E21. Grayscale information. Here, the gray-scale information storage bit string number of each pixel constituting the processed image data corresponding to the processed image E generated by the processed image generating unit 61 is a bit string of "8", and It can store 256 gray scales by storing "0" or "1" in 8 yuan. For example, the grayscale value of the pixel in which the "00000000" pixel is stored in the grayscale information storage bit string is "0 (zero)", and the grayscale value of the pixel in which the "11111111" pixel is stored in the grayscale information storage bit string is " 255".

Moreover, the processed image F shown in FIG. 8B is processed by the second defect detection algorithm. And the generated processed image is composed of a grayscale information storage bit string, that is, the second defective pixel group F21 extracted by the processing of the second defect detection algorithm stores the representation corresponding to the feature value. The gray scale information of the gray scale value, and the gray scale information indicating the gray scale value of zero is stored for the remaining pixel group F22 except the second defective pixel group F21. Here, the gray scale information storage bit string number of each pixel constituting the processed image data corresponding to the processed image F generated by the processed image generating unit 61 is a bit string of "8", and It can store 256 gray scales by storing "0" or "1" in 8 yuan. For example, the grayscale value of the pixel in which the "00000000" pixel is stored in the grayscale information storage bit string is "0 (zero)", and the grayscale value of the pixel in which the "11111111" pixel is stored in the grayscale information storage bit string is " 255".

The processed image generating unit 61 combines the processed image E generated by the processing by the first defect detecting algorithm with the processed image F generated by the second defect detecting algorithm, and is generated as shown in FIG. 8C. Process image G. The processed image G shown in FIG. 8C is composed of the first defective pixel group G21 based on the processed image E, the second defective pixel group G22 based on the processed image F, and the first defective pixel group G21 and the second defective pixel. The remaining pixel group G23 other than the group G22 is configured.

In the example shown in FIG. 8C, in the processed image G generated by the processed image generating unit 61, the grayscale information storage bit string G31 of the pixel located at the center of the first defective pixel group G21 is stored with a representation. "11111111" having a grayscale value of "255", and "01111111" indicating a grayscale value of "128" is stored in the grayscale information storage bit string G32 of the pixel located at the center of the second defective pixel group G22. The grayscale information storage bit string G33 of each pixel of the remaining pixel group G23 stores "00000000" indicating that the grayscale value is "0 (zero)".

In the present embodiment, the processed image generating unit 61 acquires defect information on the defect in the sheet-like formed body 2, that is, the defect information, based on the processed image data corresponding to the processed image G shown in FIG. 8C. The processed image G used by the processed image generating unit 61 to acquire the defect information is processed by processing the processed image E and the processed image F by processing a plurality of (two types) defect detecting algorithms having different defect detecting capabilities. Produced, therefore, can represent a sheet The defect type information of the type of the defect in the molded body 2 is included in the defect information acquired by the processed image generating unit 61 based on the processed image data corresponding to the processed image G. Specifically, the processed image generating unit 61 can detect whether the grayscale information stored in the grayscale information storage bit string constituting the processed image G is detected by using any of the plurality of defect detecting algorithms. The method performs the processing to calculate the gray scale information corresponding to the feature value, and obtains the defect information including the defect type information.

The processed image data corresponding to the processed image G output from the processed image generating unit 61 is input to the analysis image generating unit 62. 9A and 9B are views showing an example of an analysis image generated by the image processing device 6.

The analysis image generation unit 62 of the image processing device 6 pairs the first defective pixel group G21, the second defective pixel group G22, and the remaining pixel group G23 that constitute the processed image G generated by the processed image generating unit 61. Each gray scale information storage bit string is additionally attached with the defect information storage bit string storing the defect information, and an analysis image H as shown in FIG. 9A is generated. The analysis image H is composed of a parsing bit string obtained by adding the defect information storage bit string to the grayscale information storage bit string. The analysis image generation unit 62 outputs analysis image data corresponding to the generated analysis image H.

The analysis image H shown in FIG. 9A is an image composed of pixels from one end in the X direction (the left end of the analysis image H in FIG. 9A) toward the other end (the image for analysis in FIG. 9A). The right end of H is arranged in the order of 0, 1, 2, ..., W-2, and W-1, and W pixels arranged in the X direction and one end in the Y direction (the upper end of the analysis image H in Fig. 9A) To the other end (the lower end of the analysis image H in Fig. 9A), H pixels arranged in the Y direction are positioned in the order of 0, 1, 2, ..., H-2, and H-1.

In FIG. 9A, the analysis image H has a position from the one end in the X direction (the X coordinate value) is "8", and the pixel from the one end in the Y direction (the Y coordinate value) is "6" becomes the maximum brightness value. The first defective pixel group H21, the position from the one end in the X direction (the X coordinate value) is "W-5", and the pixel from the one end in the Y direction (the Y coordinate value) is "3" becomes the maximum brightness value. The second defective pixel group H22 and the remaining pixel group H23 other than the first defective pixel group H21 and the second defective pixel group H22.

In the analysis image H, the first defective pixel group H21 is a pixel group corresponding to the first defective pixel group G21 in the processed image G generated by the processed image generating unit 61, and the second defective pixel group H22 is associated with The pixel group corresponding to the second defective pixel group G22 in the processed image G generated by the image generating unit 61 is processed, and the remaining pixel group H23 is the remaining pixel group in the processed image G generated by the processed image generating unit 61. The pixel group corresponding to G23.

As shown in FIG. 9B, in the analysis image H, each pixel of the first defective pixel group H21 is composed of the analysis bit string H31, and the analysis bit string H31 is the first pair of the processed image G. The gray-scale information storage bit string H311 corresponding to the gray-scale information storage bit string G31 of the defective pixel group G21 is additionally added with the defect information storage bit string H312 of the defect information. The defect information storage bit string H312 of the analysis bit string H31 is, for example, a bit string having a bit number of "2", and stores "0" or "1" in two bits to indicate defect type information as a defect. News. In the example shown in FIG. 9B, in the analysis bit string H31 of the pixel located at the center of the first defective pixel group H21, the grayscale information storage bit string H311 stores a grayscale value of "255". "11111111" stores "01" in the defect information storage bit string H312. The "01" indicates that the grayscale information stored in the grayscale information storage bit string H311 is used by using the first defect detection algorithm and the first 2 Gray-scale information corresponding to the feature value calculated by the processing of the first defect detection algorithm in the defect detection algorithm.

Further, in the analysis image H, each pixel of the second defective pixel group H22 is composed of an analysis bit string H32 as shown in FIG. 9B, and the analysis bit string H32 is paired with the processed image G. The gray-scale information storage bit string H321 corresponding to the gray-scale information storage bit string G32 of the second defective pixel group G22 is added with a bit string formed by the defect information storage bit string H322 storing the defect information. The defect information storage bit string H322 of the analysis bit string H32 is, for example, a bit string having the number of bits of "2", and stores "0" or "1" in two bits to indicate defect type information as a defect. News. In the example shown in FIG. 9B, the second defective pixel is located In the analysis bit string H32 of the pixel of the group H22, "01111111" indicating that the grayscale value is "128" is stored in the grayscale information storage bit string H321, and the defect information storage bit string H322 is stored. 10", the "10" indicates that the grayscale information stored in the grayscale information storage bit string H321 is processed by using the second defect detection algorithm in the first defect detection algorithm and the second defect detection algorithm. And the calculated grayscale information corresponding to the feature value.

Further, in the analysis image H, each pixel of the remaining pixel group H23 is composed of an analysis bit string H33 as shown in FIG. 9B, and the analysis bit string H33 is paired with the remaining pixels of the processed image G. The gray-scale information storage bit string H331 corresponding to the gray-scale information storage bit string G33 of the group G23 is additionally added with the defect information storage bit string H332 of the defect information. The defect information storage bit string H332 of the analysis bit string H33 is, for example, a bit string having the number of bits of "2", and stores "0" or "1" in two bits to indicate defect type information as a defect. News. In the example shown in FIG. 9B, in the analysis bit string H33 of each pixel of the remaining pixel group H23, the grayscale information storage bit string H331 stores the grayscale value "0 (zero)". "00000000", stored in the defect information storage bit string H332 with "00", the "00" indicating the grayscale information stored in the grayscale information storage bit string H331 in the first defect detection algorithm and the second defect detection algorithm In any of the defect detection algorithms, the feature value of the predetermined threshold or more is not calculated and is "not defective."

In the above description, the defect information storage bit string H312, H322, and H332 in the analysis bit strings H31, H32, and H33 constituting the analysis image H are stored with the defect type information as the defect information. However, it is not limited to such a configuration.

Examples of the defect information stored in the defect information storage bit string other than the defect type information include position information of the defect in the sheet-shaped molded body 2 and the like. For example, when the location information of the defect is stored as the defect information, the X and Y coordinate values of the respective pixels may be stored in the defect information storage bit string H312, H322, and H332.

Analysis image corresponding to the analysis image H output from the analysis image generating unit 62 The data is input to the image analysis device 7.

Returning to Figure 2, the description continues. The image analysis device 7 provided in the defect inspection device 100 of the present embodiment analyzes the analysis image data output from the analysis image generation unit 62 of the image processing device 6 configured from the image generation device 1 The predetermined image analysis is performed using the information of the bit elements of the bit strings H31, H32, and H33, thereby detecting the defects of the sheet-like molded body 2. The image analysis device 7 includes an analysis image input unit 71, an image analysis unit 72, a control unit 73, and a display unit 74. The analysis image input unit 71 inputs the analysis image data output from the analysis image generation unit 62 of the image processing device 6.

The image analysis unit 72 analyzes the information stored in the analysis bit strings H31, H32, and H33 in the analysis image data input from the analysis image input unit 71 to generate a defect related to the defect. The position information, the defect brightness information, the defect type information, and the like are output to the control unit 73.

For example, the image analysis unit 72 converts the coordinates of the defective pixels in the analysis image H into positions on the sheet-like formed body 2, and generates defect position information indicating the position of the defect in the sheet-like formed body 2, and The generated defect position information is output to the control unit 73.

Further, the image analysis unit 72 converts the distribution of the gray scale information of the defect in the analysis image H into the luminance distribution of the defect on the sheet-like formed body 2, and generates a luminance distribution indicating the defect in the sheet-like formed body 2. The defect brightness information is output to the control unit 73.

Further, the image analyzing unit 72 converts the distribution of each defect in the analysis image H into the distribution of each defect on the sheet-like formed body 2, and produces a distribution indicating each of the defects in the sheet-like formed body 2. The defect type information is output to the control unit 73.

The control unit 73 creates a defect map indicating the defect information in the sheet-like molded body 2 based on the information related to the defect output from the image analysis unit 72, and collectively controls the analysis image input unit 71 and the image analysis unit 72. And a display unit 74. Defect map made by the control unit 73 It is displayed on the display unit 74.

[Industrial availability]

In the defect inspection apparatus 100 of the present embodiment configured as described above, the defect detection of the sheet-like molded body 2 is performed based on the two-dimensional image data of the sheet-like molded body 2 imaged by the imaging device 5, and thus, for example, A higher defect detection capability can be maintained as compared with the case of performing defect detection using one-dimensional image data obtained by the line sensor.

Further, in the defect inspection apparatus 100 of the present embodiment, the two-dimensional image data having a large amount of information output from the image pickup device 5 is converted into processed image data constituting each pixel by the gray scale information storage bit string, and further The image data for parsing of each pixel is converted into a parsing bit string formed by adding a defect information storage bit string to the grayscale information storage bit string. The image analysis device 7 detects the defect of the sheet-shaped molded body 2 by performing image analysis using the analysis image data of each pixel by the analysis bit string converted from the two-dimensional image data in this manner, and thus can detect the defect of the sheet-shaped molded body 2 The image analysis device 7 is required to increase the speed of image analysis, and the inspection efficiency can be improved.

Claims (5)

An image generating device that generates image data for inspecting a defect of a sheet-shaped formed body, and includes: a conveying portion that conveys the sheet-shaped formed body along a longitudinal direction of the sheet-shaped formed body; and a light-irradiating portion a light source that linearly extends in a width direction perpendicular to the longitudinal direction of the sheet-like formed body, and the sheet-shaped molded body is irradiated with light by the light source, and the image forming unit faces the sheet-shaped formed body conveyed by the transport unit. Performing a photographing operation to generate two-dimensional image data representing a two-dimensional image; the feature value calculating unit is configured to calculate the two-dimensional image data based on the luminance values of the pixels by one or a plurality of algorithms a feature value of each pixel; a processed image data generating unit that divides each pixel constituting the two-dimensional image data into a defective pixel whose feature value is equal to or greater than a predetermined threshold value, and the feature value does not reach the threshold value Remaining pixels, and generating processed image data, the processed image data comprising grayscale information storing grayscale information indicating grayscale values corresponding to the feature values for the defective pixels And storing a bit string, and storing a grayscale information storage bit string storing grayscale information indicating a grayscale value of zero for the remaining pixels; and a defect information acquiring unit, for each pixel according to the processed image data Obtaining defect information about a defect in the sheet-shaped formed body, and generating a defect information storage bit string storing the defect information obtained as described above; and an image data generating unit for analyzing, which generates an image for analysis for each pixel And the analysis image data includes a parsing bit string obtained by adding the defect information storage bit string to the grayscale information storage bit string of the processed image data. An image generating apparatus according to claim 1, wherein The defect information includes defect type information indicating the kind of the defect in the sheet-shaped formed body. The image generating device of claim 2, wherein the feature value calculating unit calculates the feature value by processing a plurality of algorithms, and the defect information acquiring unit stores the bit string according to the gray level information of each pixel. Whether the order information is the gray level information corresponding to the feature value calculated by the processing of any one of the above plurality of algorithm processes, and obtaining the defect information including the defect type information. A defect inspection device comprising: the image generation device according to any one of claims 1 to 3; and an image analysis device stored in the image data generation unit configured by the image generation device The information of the analysis target image data is analyzed by using the information of the bit string, and predetermined image analysis is performed to detect the defect of the sheet-shaped molded body. A defect inspection method for inspecting a defect of a sheet-shaped formed body, comprising the steps of: carrying a step of conveying a sheet-shaped formed body along a length direction of the sheet-shaped formed body; and a step of irradiating light by means of a sheet a light source extending linearly in a width direction perpendicular to the longitudinal direction, and irradiating the sheet-shaped molded body to be conveyed with light; and an imaging step of causing the sheet-shaped molded body during conveyance to perform an image capturing operation a two-dimensional image data of a dimensional image; a feature value calculation step of processing, by one or more kinds of algorithms, calculating a feature value of each pixel constituting the two-dimensional image data based on a luminance value of each pixel; a processing image data generating step of dividing each pixel constituting the two-dimensional image data into a defective pixel having the feature value equal to or greater than a predetermined threshold value, and remaining pixels having the feature value not reaching the threshold value, and generating a processing map The image data includes a grayscale information storage bit string storing grayscale information indicating a grayscale value corresponding to the feature value for the defective pixel, and storing the representation zero for the remaining pixels Gray scale information storage bit string of gray scale information of gray scale value; defect information obtaining step, which acquires defect information about defects in the sheet shaped body for each pixel according to the processed image data, and generates and stores a defect information storage bit string having the defect information obtained above; a parsing image data generating step of generating parsing image data for each pixel, wherein the parsing image data includes the above-mentioned processed image data a grayscale information storage bit string to which the parsing bit string obtained by adding the defect information storage bit string is added; and an image parsing step, which enables Configuration is stored in the analyzing the image data by the analyzing only the information bits string, the predetermined image analysis, thereby detecting defects in the sheet-like molded into the.