JP4903031B2 - Apparatus and method for inspecting appearance of sheet material having translucency - Google Patents

Description

  The present invention relates to an appearance inspection of a translucent sheet material.

In the linen supply industry, a large amount of laundry is processed. For example, in hotels, sheets and covers are washed in large quantities. Such laundry is ironed, for example, with a 3 m wide irona, then illuminated on the surface with lighting fixtures while being carried on a belt conveyor, the appearance is inspected, and then sent to a folding machine. . In the folding machine, the folding method and the taking-out place are changed and discharged based on the inspection result as to whether a defect such as a hole or a dirt is detected by the appearance inspection or is a non-defective product. The laundry such as sheets is an example of a sheet member having a light-transmitting property. Similarly, as for the sheet material having translucency such as film, glass cloth, prepreg, plastic board, paper, etc., as described in the appearance inspection apparatus described in JP-A-7-229852, the appearance inspection is performed. Is performed in the same manner, and defective products are detected.

When inspecting a translucent sheet material, it is affected by the color of the surface of the belt conveyor behind the object to be inspected. Therefore, in the appearance inspection apparatus described in Japanese Patent Application Laid-Open No. 7-229852, a plurality of single colors and different colors (for example, white, black, and The red block is placed so that it becomes the background of the sheet material. The color of the block may be appropriately selected according to the color of the test object. A plurality of cameras for detecting the colors of the plurality of blocks are provided. Thereby, the defect can be clearly revealed by the color difference, and the defect can be reliably detected.
JP-A-7-229852

  However, in the appearance inspection apparatus described in Japanese Patent Application Laid-Open No. 7-229852, a plurality of cameras must be provided. Therefore, the camera, the lens, the cable, and the image processing apparatus must prepare the number of colors of the block. There wasn't.

An object of the present invention is to so the appearance of the sheet material having a light transmitting property can be easily inspected.

  The appearance inspection system according to the present invention for inspecting the appearance of a translucent sheet material is: (a) a plurality of single-color background plates having different colors, through which the object to be inspected passes. A background plate provided over the conveyance width in a direction orthogonal to the conveyance direction of the object to be inspected, and (b) a camera incorporating a plurality of linear color sensors, wherein the plurality of linear color sensors are respectively A camera installed at a position for capturing an image on one of the plurality of background plates, and (c) when the same location in the inspection object passes through the inspection object on the plurality of background plates. An image processing apparatus for inspecting an appearance defect of the inspection object from an image of the inspection object captured by the camera;

  In the appearance inspection system, the plurality of background plates are, for example, a white background plate and a black background plate. Preferably, the image processing apparatus captures white calibration data based on image data obtained by imaging the white background plate when the black background plate is imaged and it is determined that there is no inspection object. Create

  In the appearance inspection system, the image processing apparatus determines that there is a tear if the image at the same location is black on the black background block and white on the image on the white background block. Further, the image processing apparatus determines that there is black stain if the image at the same location is black on the black background block and black on the white background block. In addition, the image processing device may be configured such that a part of the object to be inspected is provided if the image at the same location has higher brightness than the surrounding color on the black background block and the same brightness as the surrounding color on the white background block. Is determined to have been multiplexed.

  In the first appearance inspection method for inspecting the appearance of a translucent sheet material according to the present invention, (a) a plurality of single-color background plates having different colors are orthogonal to the conveyance direction of the object to be inspected. (B) a camera incorporating a plurality of linear color sensors, and each of the plurality of linear color sensors displays an image of an object to be inspected on one of the plurality of background plates. Install it at the position to capture images. (C) When the same location in the inspection object passes over each of the plurality of background plates, an image of the inspection object is captured by the camera, and (d) in the image of the inspection object, A defect in the appearance of the inspection object is detected by comparing images of the same part.

  In the second appearance inspection method for inspecting the appearance of a translucent sheet material according to the present invention, (a) an image of the inspection object is taken on a background plate having a color different from that of the inspection object; (B) If there is a portion with higher brightness than the single portion of the sheet material in the image of the inspection object, it is determined that the sheet material is a multiple portion.

Since one camera having a plurality of fields of view is used, high-precision appearance inspection can be provided at low cost.
By applying a background of a color different from that of the inspection object, it is possible to inspect the bending failure of the inspection object. Further, it is possible to detect a frame wrapping defect by applying a background of a color different from that of the inspection object.
By making one of the background plates white, automatic white calibration is possible.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.

  In the present invention, a defect in the appearance of a light-transmitting inspection object such as a sheet is detected.

  FIG. 1 shows one row of the overall configuration of the washing line. A laundry such as a sheet washed by a washing machine is spread in a semi-dry state, put into the irona 10, and ironed and dried. The dried laundry is transported to the folding machine 14 by the transport device 12. In the case of a sheet, it is conveyed as a single spread cloth. The imaging device 16 images the laundry to be moved on the conveyor belt by the conveyor device 12. The image processing device 18 detects an abnormal appearance (breaking, dirt, etc.) based on the video signal received from the imaging device 16. In the folding machine 14, the folding method and the taking-out place are changed based on the inspection result of the image processing device 18, and the paper is discharged. The irona 10 and the conveying device 12 have a width of 3 m, for example. In order to effectively use such a wide width, the transport device 12 transports a large number of laundry items in parallel.

  The sheet defect will be described in more detail. The sheet itself is broken and dirty, and the stain is further colored and black. Although it is not a defect of the sheet itself, the shape may become abnormal or broken during the conveyance. In this case, there is a problem that the folding machine cannot be folded normally or the sheets are clogged in the middle.

  Both tears and dirt are defective sheets, but the dirt is rewashed or blotted, whereas the tears are discarded or repaired, so it is desirable that they be discharged separately. However, the conventional inspection apparatus using only the black background plate has a problem that it cannot be distinguished whether it is torn or black.

  FIG. 2 schematically shows an example of an appearance inspection apparatus. A background block 22 of the first color and a background block 24 of the second color are arranged in parallel on a part of the conveyance path of the conveyance device 12 in a direction orthogonal to the arrangement direction over the conveyance width (for example, 3 m) of the conveyance device 12. The The first color is black and the second color is white. The black and white background blocks 22 and 24 are manufactured by, for example, black and white coating of an iron plate. The laundry 20 that is ironed out of the irona 10 passes over the two-color background blocks 22 and 24. The white light sources 30a and 30b are installed forward and backward in the transport direction at positions where the inspection objects on the background blocks 22 and 24 are illuminated, and illuminate the inspection objects brightly. Further, a multi-plate line camera (two-plate line camera 32 in this example) is installed above the two-color background blocks 22 and 24. The two-plate camera 32 includes two light receiving elements 34a and 34b. The light receiving elements 34a and 34b are CCD linear sensors. The positions of the light receiving elements 34a and 34b are images of the laundry 20 on the first color background block 24 via the lens 36, and the light receiving element 34b is the laundry on the second color background block 22 via the lens 36. It is adjusted so that 20 is imaged.

  A multi-plate line camera will be described. Usually, the line camera has a single field of view, and two line cameras are required to realize a configuration in which the background blocks 22 and 24 of two colors are observed. When there are two line cameras, two lenses, cables, and an image processing device are required. In order to obtain a 3 m field of view width, it is necessary to install the camera at a considerably high position. The position and orientation of the camera installed at the high position are adjusted, and the relative positional relationship between the fields of view of the two cameras is accurately determined. It is very difficult to match.

  Therefore, a two-plate line camera 32 was created for appearance inspection. Here, two linear sensors (light receiving elements) 34 a and 34 b are arranged in parallel in one camera 32. That is, one camera 32 has a plurality of fields of view. When three color background blocks are used, three linear sensors are arranged in parallel in one camera.

  In the case of a two-plate camera, since the relative positional relationship between the two linear sensors has been adjusted in advance at the camera manufacturing stage, the camera is adjusted so that the field of view of the linear sensor that captures the background block fits in one background block. Just do it. On the other hand, in the case of the two-camera type, it is not sufficient to adjust the field of view of each camera so that it can be accommodated in each background block. , Y axis and θ direction must be adjusted precisely. In addition, since there is an individual difference in the lens focal length, it is necessary to adjust the height direction (Z direction) of the camera in order to adjust the imaging magnification.

  In the camera 32, analog signals from the image pickup devices 34a and 34b are amplified by an amplifier circuit, and AD conversion of the amplified signal is performed by an AD converter. Then, the hardware logic circuit calibrates the obtained digital signal (raw data) using the whiteboard calibration data stored in advance, and outputs an 8-bit digital signal.

  The background blocks 22 and 24 have a single color. Here, the background blocks 22 and 24 are black and white, respectively. If necessary, three or more background blocks may be used. In this case, the camera incorporates three or more linear sensors.

  As shown in FIG. 3, the image processing apparatus has a configuration similar to that of a normal computer. The CPU 40 for controlling the whole includes an input device 42 such as a keyboard for inputting commands, a display device 44 for displaying data, a storage device 46 such as a hard disk drive, and an interface 48 for signal input / output with the camera 32. . The storage device 46 stores image data input from the camera, white calibration data, an image processing program for appearance inspection, and the like.

  The operation of the appearance inspection apparatus described above will be described. In the transport device 12, the laundry 20 passes over the black background block 22 and then passes over the white background block 24. The laundry 20 is irradiated with white light sources 30a and 30b. Since the light receiving elements 34a and 34b are arranged at regular intervals, the time required for the portion on the black background block 22 to reach the black background block 24 by performing a reverse calculation using the moving speed of the laundry 20 I understand. Therefore, the light receiving elements 34a and 34b can image the same portion of the laundry 20 with a time difference. Next, by comparing the image on the black background block 22 and the image on the white background block 22 with respect to the same portion of the inspection object, the presence / absence and type of the defect can be determined.

  In this appearance inspection apparatus, it is possible to inspect four kinds of defects (tear, black stain, color stain, crease) as shown in Table 1 for a light-transmitting inspection object (sheet or the like). (Breaks, that is, the holes are in the state of highest translucency.) Each example will be described below.

Table 1 Defect detection

  When the laundry 20 is torn, since the background block is exposed at the torn portion, the laundry 20 is imaged by the light receiving element 34a on the black background block 22 with the tear black. The Next, the laundry 20 is imaged on the light receiving element 34b on the white background block 24 with white tearing. It is possible to recognize that the laundry 20 has been torn by imaging the same part of the laundry 20 with a time difference by the light receiving elements 34a and 34b. (Here, the colors of the image are simply expressed as black and white for the sake of simplicity, but in reality, even if the tears are on the black background block, the tears that are half transparent may only be dark and may not be black. Also, even if a tear occurs on the white background block, it may appear partially black due to the surrounding shadow, etc. Therefore, the tear is detected by the relative difference between the black background block and the white background block.)

  Further, when there is black dirt on the laundry 20, first, the black background block 22 is black and is picked up by the light receiving element 34 a. Next, even in the white background block 24, the dirt is imaged black. It can be recognized that if the light receiving elements 34a and 34b pick up images of the same portion of the laundry in a time lag and are picked up as black at the same portion, it is black dirt.

  Further, when there is a color stain on the laundry 20, the color difference may be detected because it is different from the surrounding colors. However, since the white background block 24 can capture an image with higher contrast, the white background block It is preferable to use the image on 24.

  Furthermore, it has been found that when the laundry is inspected by this appearance inspection apparatus, it is possible to effectively detect the sheet folding failure from the luminance information. FIG. 4 shows an example of the folding failure, and a part of the sheet is folded on the lower left side. Sheets for which a defective shape could not be detected are sent to the folding machine as they are, so that parts other than the normal folding are folded or the folding machine is clogged. It is therefore desirable to be able to inspect such defects. Considering the conventional technology from this point of view, conventionally, a plurality of photoelectric sensors are arranged within the width of the transport device, and the state of the end of the sheet is estimated based on the obtained discrete position information. It was broken. However, this method has a problem that only a defective shape at a position where the photoelectric switch can be detected can be detected. A method of detecting a shape defect based on the outer shape of the entire image of the inspection object instead of the photoelectric sensor is also conceivable. In this case, there is no problem due to the discrete positions of the photoelectric sensor, but even if a defect in the outer shape can be detected, a broken defect cannot be detected. When the corner as shown in FIG. 5 (a) is broken, it can be detected because the outer shape is different from the rectangle, but the outer shape is rectangular even if it is bent as shown in FIG. 5 (b). If you are doing, you can not detect defects in the outer shape.

  When the laundry 20 partially overlaps, the lower cloth serves as a white background block in the overlapping portion on the black background block 22, and a high-intensity image is captured. On the other hand, in the non-overlapping part, the dark color of the black background 22 is transmitted through the laundry, resulting in a low brightness image compared to the two-layered part. An example is shown on the lower side of FIG. Subsequently, when the laundry 20 passes over the white background block 24, the background block 24 has already been seen through to create a high-luminance image. It becomes a picture. Accordingly, it is possible to identify whether the laundry is a laundry that is partially overlapped with thin dirt or a laundry that has an overlapping portion. If the item to be washed is a single sheet, it can be detected that the sheet is not broken.

  FIG. 6 shows a flowchart of whiteboard calibration data generation by the image processing apparatus 18. White plate calibration is stable because it eliminates unevenness due to uneven illumination, lens shade (periphery is dark), variation in CCD sensitivity, etc., and light intensity fluctuations (temporal and positional) mainly due to illumination. It means that a white plate that can be regarded as uniform is imaged and calibrated using the imaged data.

  First, the image on the black background block 22 is picked up by the camera 32 and image data is acquired (S10). Next, the presence or absence of a sheet is checked based on the image data (S12). If there is no sheet, the process returns to step S10. Here, if there is a bright portion in the imaging data in the black background block 22, it is determined that there is a sheet, and if it is not, it is determined that there is no sheet. Actually, a method for determining whether the number of pixels equal to or greater than the set threshold value is greater than the set number of pixels in consideration of dirt or dust on the black background block, a method for determining based on an average value, or the like can be considered. Alternatively, the determination may be made using only one line of data, or using a plurality of lines of data.

  When the sheets are detected, next, the image on the white background block 24 is captured by the camera 32 to acquire image data (S14). Due to the distance between the white background visual field and the black background visual field, a time delay occurs. For this reason, even if it is determined that there is no sheet in the black background block, there may still be a sheet on the white background block 24. Therefore, it is necessary to adjust the time until shooting with the white background field of view by idle reading. is there. Further, data may be acquired not only by one line but also by a plurality of lines.

  Next, white plate calibration data is calculated (S16). Here, noise is removed by acquiring and averaging a plurality of lines. Alternatively, the influence of dust or dirt may be reduced by smoothing in the pixel arrangement direction (lateral direction). Next, the whiteboard calibration data is updated with the calculated whiteboard calibration data (S18).

  This white board calibration data is sent to the camera 32 and stored in the camera 32. On the right side of FIG. 4, a processing procedure for image data in the camera 32 is shown. The inspected object is imaged by the imaging elements 34a and 34b, the obtained analog signal is amplified by an amplifier, and then AD conversion (12 bits) is performed to obtain a digital signal (raw data) (S20). The raw data is calibrated using the white board calibration data by the hardware logic circuit, and an 8-bit digital signal is output (S22). Then, the 8-bit digital signal is output to the image processing device 18 (S24). The image processing apparatus 18 inputs this and determines the presence and type of appearance abnormality. The photographing data processing described above has a drawback that calibration data must be transmitted to the camera 32. However, since white plate data and the like can be grasped in the image processing apparatus 18, it is possible to detect a warning such as deterioration of illumination or white plate dirt. Etc. can be issued.

  Other schemes are possible. For example, the image processing apparatus determines only the presence or absence of a sheet and gives a white board calibration command to the camera. Then, all processing including white plate calibration data calculation is performed in the camera. This method reduces the burden on the image processing apparatus and does not require data transfer, but cannot be managed as a black box when viewed from the image processing apparatus. Further, it is necessary to provide a complicated arithmetic processing circuit on the camera side. In another method, only analog or digital raw data is output from the camera, and all processing is performed by hardware or software on the image processing apparatus side. In this case, post-calibration data output means output to processing logic (hardware, software). In this method, since the output data from the camera is raw data, if there are not many gradations, digits are lost and the SN ratio is deteriorated. In addition, the burden on the image processing apparatus increases.

  When detecting the overlap of translucent sheet material (inspection object), as described above, on the background plate of a color (for example, black) different from the inspection object (for example, white), Take an image of an object. Then, if there is a portion with higher brightness than the single portion of the sheet material in the image of the inspection object, it is determined that the sheet material is a multiplexed portion. Here, if the portion where the cloth is doubled and the portion where the cloth is overlapped are separated from the inspected image from the luminance information, the inspection can be performed for each area. For example, when the laundry is the frame wrap shown in FIG. 6, the inspection sensitivity and the like are changed between the inner area and the outer area of the frame. For the single portion (for example, the above-described inner area), various appearance abnormalities are determined by the above-described method. For the double part, only the dirt inspection is performed because the tear inspection is impossible.

  Note that the current area sensor cannot be employed as an image sensor for the camera 32. However, an area sensor capable of partial scanning is known, and such an area sensor can read out faster by partial scanning. Therefore, for example, if an area sensor capable of imaging a 3 m width can be used, such an area sensor can be installed in the camera 32 instead of the linear sensor. When the inspection object is small, a camera provided with an area sensor may be used. Further, when displaying the entire image for detecting the overlap of the inspection objects, if the inspection object is small, a camera equipped with an area sensor can be used.

  Needless to say, the present invention can be similarly applied to an appearance inspection of a light-transmitting sheet material such as a film, a glass cloth, a prepreg, a plastic board, and paper in addition to a laundry such as a sheet.

The figure which shows the whole structure of a washing line The figure which shows the composition of the appearance inspection system Illustration of an example of an image processing device The figure which shows one example of a bending defect Diagram for explaining broken defect detection The figure which shows an example of frame wrapping Whiteboard calibration data generation flowchart

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 Irona, 12 Conveying device, 16 Imaging device, 18 Image processing device, 20 Washing object, 22 First color background block, 24 Second color background block 24, 30a, 30b Light source, 32 Two-plate line camera, 34a, 34b Light receiving element (linear sensor), 36 lens.

Claims (8)

An appearance inspection system for inspecting the appearance of a translucent sheet material to be inspected,
The inspection object which is conveyed passes thereon a plurality of colors from each other is different single color background plate of, placed across the conveying width in a direction perpendicular to the conveying direction of the object to be inspected, a plurality With a background board,
A multi-plate camera incorporating a plurality of linear color sensors and a lens common to the plurality of linear color sensors, the plurality of linear color sensors having their relative positions fixed to each other, color sensor, wherein each via a common lens passes over the different one background plate of the plurality of background plates are installed an image of the object on the imaging position capable multi-plate camera and,
When the same position in the inspection object passes over the plurality of background plates, by comparing the image of the same portion of the multi-plate the inspection object captured by the linear color sensor camera an image processing apparatus for inspecting defects of appearance of the object to be inspected,
Appearance inspection system equipped with.   The appearance inspection system according to claim 1, wherein the plurality of background plates are a white background plate and a black background plate. The image processing apparatus according to an image of the same portion is a black on a background plate of the black color, if white background board of the white color, and judging the break is present Item 2. The appearance inspection system according to item 2. The image processing apparatus includes a feature that an image of the same portion is a black on a background plate of the black color, if black background board of the white color, it is determined that black stain is present The visual inspection system according to claim 2. The image processing apparatus, wherein a high brightness than the surrounding color image at the same position is on a background plate of the black color, if the same brightness as the surrounding colors on the background plate of the white color, the object 3. The appearance inspection system according to claim 2, wherein it is determined that a part of the inspection object overlaps and is multiplexed. The image processing apparatus, when the object to be inspected by imaging the background plate above the black is determined not exist, to calibrate the white data on the basis of the image data obtained by imaging the white background plate The appearance inspection system according to claim 2, wherein: An appearance inspection method for inspecting the appearance of a translucent sheet material to be inspected,
A plurality of single-color background plate of a color different from each other, placed across the conveying width in a direction perpendicular to the conveying direction of the object to be inspected,
A multi-plate camera incorporating a plurality of linear color sensors and a lens common to the plurality of linear color sensors , and fixing the relative positions of the plurality of linear color sensors to each other, Each of the plurality of background plates through the lens of each of the plurality of background plates installed on a position to capture an image of the inspection object passing over one different background plate ,
When the same position in the object to be inspected passes over each of the plurality of background plates, the image captured for the same position of the inspection object by the linear color sensor of the multi-plate type camera,
Wherein by comparing the image of the same position in the image of the object, detecting defects in the appearance of the object to be inspected,
Appearance inspection method. An appearance inspection method for inspecting the appearance of a translucent sheet material to be inspected,
A white background plate and a black background plate are installed across the conveyance width in a direction perpendicular to the conveyance direction of the inspection object,
A two-plate camera incorporating two linear color sensors and a lens common to the two linear color sensors, the relative positions of the two linear color sensors are fixed to each other, and each linear color sensor is the common Through the lens of each of the white background plate and the black background plate are installed at a position to capture an image of the inspection object passing over one different background plate,
When the same location in the inspection object passes over each of the two background plates, the linear color sensor of the 2-plate camera captures an image of the same location of the inspection object,
If the image of the same location is higher in luminance than the surrounding color on the black background plate and has the same luminance as the surrounding color on the white background plate, a part of the inspection object overlaps. Judge that it is multiplexed,
Appearance inspection method.

Images