JP4917364B2 - Inspection apparatus and inspection method - Google Patents

Description

  The present invention relates to a defect inspection method and a defect inspection apparatus for determining the quality of an arrangement state of members in a cylindrical light diffuser in which two or more members having different light transmittances are arranged.

  A cylindrical light diffuser in which two or more kinds of members having different transmittances such as a cigarette filter are alternately arranged is used as an inspection object, and this is inspected. When inspecting such an inspection target, in general, the boundary positions of both ends of the inspection target and each member are detected based on a waveform from an image obtained by imaging the transmitted light with a line CCD (Charge Coupled Devices) camera. The total length, poor alignment, presence / absence of gaps between members, end member length, etc. are inspected.

In the cigarette filter inspection apparatus, for example, as shown in FIG. 12 according to Patent Document 1, cigarette filters 1 are conveyed one by one in the radial direction by a drum 9, and the drum 9 has a diameter smaller than the diameter of the filter 1, A slit 32a longer than the filter 1 is formed in the axial direction. In order to illuminate the filter 1 through the slit 32a, the end face 33a of the light guide 33 is arranged in a line at the inspection position of the filter 1. The light guide 33 guides light emitted from the light source 34. A line CCD camera 3 is disposed in the optical path of the transmitted light of the filter 1, and the line CCD is interposed between the end face 33 a of the light guide 33 and the slit 32 a by light passing between both ends of the filter 1 and the slit 32 a. A dimming member 36 for preventing the light receiving element of the camera 3 from being saturated is arranged so as to adjust the positions of both ends of the slit 32a in the axial direction.
The image of the transmitted light of the cigarette filter 1 photographed by the line CCD camera 3 is stored by the image processing apparatus 10, and the stored image is processed by the image processing apparatus 10, whereby the cigarette filter 1 is inspected.

As shown in FIG. 13A, the filter 1 has a structure in which fiber bundles 1b not containing activated carbon and fiber bundles 1a containing activated carbon are alternately arranged. The fiber bundle 1b containing no activated carbon and the fiber bundle 1a containing activated carbon are members having different transmittances. here,
Of the fiber bundles 1a and 1b, the fiber bundle 1a, which is the member with the lower transmittance, is called a dark member, and the fiber bundle 1b, the member with the higher transmittance, is called the bright member. Hereinafter, description will be made using the terms dark member and bright member.

In such a configuration, when the light transmitted through the filter 1 is imaged, the gradation level of the waveform is detected as a waveform gradation level as shown in FIG. 13B, for example. Based on the detected gradation level, the length of the entire length of the filter 1 is determined by a threshold 41 between the level of light that travels outside both ends of the filter 1 and the level of transmitted light of the fiber bundle 1b that does not contain activated carbon. Acquire and inspect whether the acquired total length is correct.
Further, the boundary between the two fiber bundles 1a and 1b is acquired by the threshold value 42 between the gradation levels of the transmitted light of the fiber bundle 1b not containing activated carbon and the fiber bundle 1a containing activated carbon. It is possible to check whether the lengths of the fiber bundles 1a and 1b are correct and whether the two fiber bundles 1a and 1b are alternately arranged. Moreover, the end member length which is the length of the fiber bundle 1a which is a member which becomes the left end and right end of the filter 1 is also acquired, and the acquired end member length is inspected.

  Further, for example, in Patent Document 2, two threshold values for detecting both ends of the light diffuser and a threshold value for identifying two members having different light transmittances are set in advance. From the result of binarization, the positions of both ends and the space between the members are detected, and the timing at which the light diffuser passes is extracted. Further, the position and arrangement state of each member are determined from the result of binarization at the member identification threshold value, and whether or not the two members of the light diffuser are alternately arranged is determined. Yes.

In the inspection apparatus according to Patent Document 1 or Patent Document 2, the image data of the cylindrical light diffuser is line-shaped image data from the line CCD camera 3. Since the line CCD camera 3 captures an image of the cigarette filter 1 that is a cylindrical light diffuser while it is being transported, the thickness of the cylindrical light diffuser that is captured by the line CCD camera 3 is changed by the transport. Further, the light is blocked by the drum 9, and the light can only pass through the slit 32 a portion of the drum 9. Therefore, as the drum 9 rotates, the area where the line CCD camera 3 captures the slit 32a changes, and the intensity of transmitted light passing through the slit 32a also changes. For this reason, the intensity of the image data as a line-like signal also changes during conveyance. For this reason, in the conventional apparatus, in the line-shaped image data obtained in the sub-scanning direction, only the line-shaped image data having the maximum gradation level of the image data is used, and the cylindrical light diffuser is used. We are inspecting.
However, when detection is performed using only one line of image data having the maximum gradation level in the sub-scanning direction, noise is caused by some influence on the image data of one line having the maximum level of image data. When mixed, normal inspection cannot be performed even if the image data of other lines are not mixed with noise.
In addition, since the portion of the cylindrical light diffuser in the sub-scanning direction in which the gradation level of the image data in the sub-scanning direction becomes maximum fluctuates every time, a plurality of cylindrical light diffusers are subjected to the same portion in the sub-scanning direction. Cannot inspect by image data.
In addition, since the portion of the cylindrical light diffuser in the sub-scanning direction in which the gradation level of the image data in the sub-scanning direction is maximized is different for each of the plurality of cylindrical light diffusers to be inspected, an appropriate threshold value is set. It is different for each cylindrical light diffuser, and it is difficult to always maintain an appropriate threshold value. In addition, it becomes difficult to compare image data of a plurality of cylindrical light diffusers.
JP-A-5-203588 JP-A-4-03719

  The present invention has been made in view of such circumstances, and the object thereof is to stabilize the member arrangement state of the cylindrical light diffuser in which two members having different light transmittances and lengths are alternately arranged. An object of the present invention is to provide an inspection apparatus that calculates image data that can be inspected.

  The present invention has been made to solve the above-described problems, and the invention according to claim 1 is directed to conveying the longitudinal light diffuser in the sub-scanning direction perpendicular to the longitudinal direction to thereby extend the light diffuser in the longitudinal direction. An inspection device for a light diffusing body that inspects in a main scanning direction parallel to a direction, wherein the light diffusing body is irradiated with light in a scanning direction, and is arranged to receive light emitted from the light source means The transmitted light, which is light emitted from the light source means and transmitted through the light diffuser, is photographed every time the light diffuser is transported in the sub-scanning direction, and the image data is arranged as pixels. An image capturing means for outputting, and pixels having a gradation level of image data in the sub-scanning direction output from the image capturing means having a predetermined threshold value or more are detected in the main scanning direction from the image data, and the detected main scanning is detected. Based on the direction pixels A feature amount extracting means for calculating each image data; and a range specifying means for extracting a range of image data in the sub-scanning direction from the image data in the sub-scanning direction based on the feature amount calculated by the feature amount extracting means; And an image extracting means for generating image data for inspection by adding the image data in the sub-scanning direction based on the range of the image data in the sub-scanning direction extracted by the range specifying means. It is.

According to a second aspect of the present invention, in the inspection apparatus according to the first aspect, the target pixel is shifted from one end of the image data for inspection generated by the image extraction means , and the maximum value of the gradation from the one end to the target pixel is detected. together determine the peak level is said to a threshold value obtained by multiplying a certain ratio to the peak level, first the pixels less than the threshold value for detecting the pixel positions arranged consecutively in a position corresponding to one end of the previous SL light diffuser the detection means, the image extracting unit will shift the target pixel from the other end of the image data for inspection is generated, with determining the peak level which is the maximum value of the gradient to the target pixel from the other end, the peak the value obtained by multiplying a certain ratio to a level as a threshold value, a second detecting means for pixels less than the threshold value for detecting the pixel positions arranged consecutively in a position corresponding to the other end of the prior SL light diffuser, the first The main scanning direction range corresponding to the light diffuser portion of the image data for inspection is larger than the range between the one end position detected by the output means and the other end position detected by the second detection means. Main scanning direction range specifying means for specifying the scanning direction range.
For example, based on the image data for inspection generated by the image extraction means, the peak level which is the maximum value of the gradation of the pixel is updated, and a value obtained by multiplying the peak level by a certain ratio is set as a threshold value. First detecting means for detecting one end position corresponding to one end of the light diffusing body in the image data for the previous inspection by searching a pixel position in which the pixels that are less than one are consecutive from one end in the main scanning direction of the image data The peak level which is the maximum value of the gradation of the pixel is updated, and a value obtained by multiplying the peak level by a certain ratio is set as a threshold value, and the pixel position where pixels which are less than the threshold value continue is the main scan of the image data. Second detection means for detecting the other end position corresponding to the other end of the light diffuser in the image data for the previous inspection by searching from the other end in the direction, and the first detection. Main scanning that is a range in the main scanning direction corresponding to the portion of the light diffuser of the image data for inspection from the range between the one end position detected by the means and the position of the other end detected by the second detection means Main scanning direction range specifying means for specifying the direction range.

According to a third aspect of the present invention, in the inspection apparatus according to the second aspect, the boundary image data for extracting the inspection image data generated by the image extracting means based on the main scanning direction range specified by the main scanning direction range specifying means. An extraction means; and a boundary threshold calculation means for calculating a threshold value by calculating a maximum value and a minimum value of the image data extracted by the boundary image data extraction means, and multiplying a difference between the calculated maximum value and minimum value by a fixed ratio And boundary position specifying means for specifying the boundary position of the light diffuser member based on the image data extracted by the boundary image data extracting means and the threshold value calculated by the boundary threshold value calculating means.

According to a fourth aspect of the present invention, there is provided an inspection apparatus for a light diffuser that conveys a longitudinal light diffuser in a sub-scanning direction orthogonal to the longitudinal direction and inspects the light diffuser in a main scanning direction parallel to the longitudinal direction. The light source means irradiates the light diffusing body with light in the scanning direction, and the photographing means is arranged to receive the light emitted from the light source means, and is irradiated from the light source means. The transmitted light, which is light transmitted through the light diffuser, is photographed every time the light diffuser is transported in the sub-scanning direction, and is output as image data in which pixels are arranged in a line shape. Pixels in which the gradation of the image data in the sub-scanning direction output from the photographing unit is equal to or greater than a predetermined threshold value are detected in the main scanning direction from the image data, and the feature amount is based on the detected pixels in the main scanning direction For each image data Then, the range specifying unit extracts a range of image data in the sub-scanning direction from the image data in the sub-scanning direction based on the feature amount calculated by the feature amount extracting unit, and the image extracting unit extracts the range by the range specifying unit. Based on the range of image data in the sub-scanning direction, the image data is added in the sub-scanning direction to generate image data for inspection.

  According to the present invention, there is provided an inspection apparatus for calculating image data capable of stably performing quality determination of a member arrangement state of a cylindrical light diffuser in which two or more members having different light transmittances are arranged. There is an effect that can be done.

  Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a configuration diagram showing the configuration of a cylindrical light diffuser such as a tobacco filter. A cylindrical light diffuser (1), which is a cylindrical long light diffuser such as a cigarette filter, is formed by alternately arranging two members 1a and 1b having different light transmittances. For example, as shown in FIG. 1, the cylindrical light diffuser (1) has a dark member 1b (1), a bright member 1a (1), a dark member 1b (2), and a bright member 1a (2) from the left in the longitudinal direction. , And the dark member 1b (3).

FIG. 2 is a diagram illustrating a configuration of the inspection apparatus. Each device will be described below.
The cylindrical light diffuser (1) is a longitudinal cylindrical light diffuser of the member 1 shown in FIG. 1, and is in a direction (sub-scanning direction) perpendicular to the longitudinal direction of the cylindrical light diffuser (1). It is conveyed by a conveyance drum (9) in a certain moving direction (1c). The illumination device (2) is disposed, for example, on the lower side of the cylindrical light diffuser (1), and illuminates light elongated in the longitudinal direction in accordance with the longitudinal direction of the cylindrical light diffuser (1).
The conveyance drum (9) is provided with a groove and a slit for holding the tobacco filter (1), and rotates in a certain direction (9a), thereby moving the tobacco filter (1) in the conveyance direction (1c). Transport. Further, this transport direction is the sub-scanning direction. As the illumination device (2) installed inside the transport drum (9), for example, an illumination device using an optical fiber according to Japanese Patent Application No. 10-250716 is used.
A line sensor (3) is arrange | positioned in the position which receives the light illuminated from the illuminating device (2). The line sensor (3) is a linear sensor that is parallel to the longitudinal direction of the conveyed cylindrical light diffuser (1) and parallel to the elongated light in the longitudinal direction of the illumination device (2). Is arranged. That is, the cylindrical light diffuser (1), the line sensor (3), and the illumination device (2) are arranged in parallel to each other. The cylindrical light diffuser (1) is conveyed between the line sensor (3) and the illumination device (2).
As shown in FIG. 3, the line sensor (3) images the cylindrical light diffuser (1) as image data in which pixels are arranged in one line in the main scanning direction parallel to the longitudinal direction. The line sensor (3) outputs the detected intensity of transmitted light to the inspection apparatus (10) as analog or digital signal image data. In addition, the line sensor (3) captures image data at regular time intervals, thereby transmitting the cylindrical light diffuser (1) along with the conveyance of the cylindrical light diffuser (1) in the sub-scanning direction. The transmitted light is imaged as image data in which a plurality of pixels are arranged in one line in the main scanning direction in the sub-scanning direction.
The inspection device (10) inspects the cylindrical light diffuser (1) based on the image data output from the line sensor (3). When the image data from the line sensor (3) is input to the inspection device (10), the inspection device (10) is input as it is when it is a digital signal, and the inspection device (10) when it is an analog signal. ) A / D converts an analog signal and inputs it as a digital signal.

Here, the illuminating device (2) uses, for example, a light guide disclosed in Japanese Patent Laid-Open No. 5-203857, adjusts the emission width of the light from the illuminating device (2), and the line sensor (3) is a cylinder. What prevents the direct light outside the both ends of the light diffuser (1) from being imaged is used.
As shown in FIG. 3A, the image data detected by the line sensor (3) generally has a high gradation level of the image data corresponding to the bright member 1a, and the gradation level of the image data corresponding to the dark member 1b. The waveform becomes low.
Moreover, since the light emission width of the light from the illumination device (2) is adjusted by the light guide according to the end of the cylindrical light diffuser (1), the end of the cylindrical light diffuser (1) is adjusted. The image data corresponding to the outside does not saturate the light and has a low level of gradation, so there is no influence of transmitted light. In addition, since the light has entered the outside of the cylindrical light diffuser (1), a gradation peak due to the transmitted light is generated just outside the end of the cylindrical light diffuser (1).
This waveform has a peak when the cylindrical light diffuser (1) is directly below the imaging range of the line sensor (3), and the level of the overall gradation decreases as it goes out of the imaging range. In other words, the image data of the cylindrical light diffuser (1) always changes with the conveyance in the conveyance direction (1C).
In the embodiment, when the transmitted light intensity is high, the gradation level of the image data obtained by photographing the transmitted light is high, and conversely, when the transmitted light intensity is weak, the image data obtained by photographing the transmitted light. In the following description, it is assumed that the gradation level is low.

Next, the inspection apparatus (10) will be described. The inspection device (10) includes a feature amount extraction processing unit (4), a recording processing unit (5), a range specifying processing unit (6), an image extraction processing unit (7), and a measurement processing unit (8).
The feature amount extraction processing unit (4) detects a feature amount indicating the quality of the image data for each image data in the sub-scanning direction of the line sensor (3), and transmits the detected feature amount to the recording processing unit (5). To do. Since the feature quantity depends on the positional relationship between the line sensor (3) and the cylindrical light diffuser (1) and the cylindrical light diffuser (1) is moving, the feature quantity is sub-scanned by the line sensor (3). Different for each. Based on this feature amount, whether or not the cylindrical light diffuser (1) is in the imaging range of the line sensor (3) and whether or not the image data of the cylindrical light diffuser (1) can be determined. This determination can be performed on the image data for each sub-scan of the line sensor (3). For example, the feature amount may be the sum of the number of pixels of image data detected at a predetermined threshold value or more in each sub-scan, or detected at a predetermined threshold value or more. The total of the gradation levels of the image data may be used.

The recording processing unit (5) receives the image data from the line sensor (3), receives the feature amount from the feature amount extraction processing unit (4), and records the received image data and the feature amount in association with each other.
The range specifying processing unit (6) refers to the plurality of sub-scanning feature amounts recorded in the recording processing unit (5), and specifies the sub-scanning range that is most effective for detection based on the feature amounts. The method for specifying the sub-scanning range may be, for example, a feature amount range having a predetermined threshold value or more. Alternatively, the method for specifying the sub-scanning range is to detect a sub-scan having the maximum feature amount, and center the sub-scan having the detected maximum feature amount as a center. Scanning may be used. The range specification processing unit (6) transmits the specified range to the image extraction processing unit (7).

  The image extraction processing unit (7) extracts the image data recorded in the recording processing unit (5) based on the sub-scanning range received from the range specifying processing unit (6), and performs measurement processing from the extracted image data. The image data for use is synthesized. For example, image data for measurement processing is synthesized by extracting image data from the recording processing unit (5) based on the sub-scanning range specified by the range specifying processing unit (6), and sub-scanning the extracted image data. Find by summing in direction. The image extraction processing unit (7) sends the combined image data to the measurement processing unit (8).

  The measurement processing unit (8) performs processing for the inspection of the cylindrical light diffuser (1) using the combined image data for measurement processing received from the image extraction processing unit (7). For example, as a process for performing the inspection, first, a process for specifying a boundary between the members is performed, and a process for inspecting the cylindrical light diffuser (1) is performed based on the obtained boundary position. For example, a process for obtaining the boundary position may be performed by performing binarization with a certain threshold value or by performing a differentiation process.

The operation of the above-described embodiment will be described below with reference to the drawings. In the block diagram showing the configuration of the inspection apparatus according to the embodiment of the present invention shown in FIG. 2, a line CCD camera (3) is used as an example of the line sensor (3), and a cigarette is used as an example of the cylindrical light diffuser (1). A case where the filter (1) is inspected will be described. In this cigarette filter (1), a fiber bundle containing activated carbon having a low light transmittance and a fiber bundle not containing activated carbon having a high light transmittance are alternately arranged, and the total length is about 100 mm.
Such a cigarette filter (1) is conveyed in the sub-scanning direction by rotating the conveyance drum (9). Here, the conveyance drum (9) is provided with a groove and a slit for holding the tobacco filter (1), and the tobacco filter (1) is conveyed by rotating in a certain direction (9a). The

The slit causes the inside of the transport drum (9) to have a negative pressure, and causes the tobacco filter (1) to adhere to the transport drum (9). Moreover, light is irradiated to a tobacco filter (1) through a slit by the illuminating device (2) installed in the inside of a conveyance drum (9).
The line CCD camera (3) captures transmitted light as image data and transmits the captured image data to the inspection apparatus (10). The transmitted light is light emitted from the illumination device (2) and transmitted through the tobacco filter (1). The line CCD camera (3) has, for example, a structure in which a plurality of photographing pixels are arranged in a straight line, and each pixel has a pixel number continuously from one end of the straight line to the other end. Is shaken. For example, the pixel numbers of 1 to 2048 are continuously assigned to the photographing pixels of the line CCD camera (3) from one end of the linear shape to the opposite end. In addition, each photographing pixel of the line CCD camera (3) converts the received light as a digital value having a gradation degree of 0 to 255 according to the intensity of the received light. Accordingly, the image data of the transmitted light photographed by the line CCD camera (3) is linear image data, and the pixel numbers from 1 to 2048 are consecutive from one linear end to the opposite end. Each pixel number is assigned a digital value having a gradation of 0 to 255. In the following description, it is assumed that the linear image data is assigned pixel numbers 1 to 2048 continuously from left to right.
The inspection device (10) receives image data from the line CCD camera (3), and inspects the tobacco filter (1) based on the received image data. The inspection method performed in the inspection apparatus (10) will be described in detail below.

The inspection device (10) receives image data from the line CCD camera (3).
First, the feature quantity extraction processing unit (4) extracts the feature quantity from the image data received by the inspection apparatus (10). As the feature amount, for example, as shown in FIG. 5, the total number of pixels of the image data equal to or higher than a slice level, which is a predetermined threshold value (number of pixels to be a + b + c + d), may be used. Alternatively, as the feature amount, for example, as indicated by hatched lines in FIG. 6, the sum of the gradation levels of image data that is equal to or higher than a predetermined slice threshold value (a range that is equal to or higher than the slice level). Of the pixel data). This feature amount extraction is performed on the received image data for each sub-scan.
Here, the slice level as the threshold value may be stored as a predetermined constant value. Alternatively, the peak level of the received image data may be detected, and the peak value of the detected gradation level may be multiplied by a predetermined ratio. For example, 30% may be stored as a predetermined fixed ratio, and a value of 30% of the detected peak level value may be used as the slice level.

Next, the recording processing unit (5) associates the image data received by the inspection device (10) with the feature amount extracted by the feature amount extraction processing unit (4) for the image data received by the inspection device (10). Record. For example, as shown in FIG. 7, the sub-scan number for identifying each sub-scan, the received image data, and the feature amount extracted from the received image data are recorded in association with each other. Here, the sub-scan number is a sequential number assigned in the order of transmission from the line CCD camera (3), for example.
In FIG. 7, the image data is displayed in the form of a waveform, but the recorded image data is a plurality of pixels arranged in a straight line, and each pixel is, for example, from left to right. The pixel numbers are consecutively assigned from 1 to 2048, and 256 gradation digital data represented by levels from 0 to 255 are stored in the respective pixels. Further, for example, the feature amount is obtained by adding the gradation values of all the pixels having a gradation level of 20 or more for the image data corresponding to one sub-scan number in the main scanning direction for all the pixel numbers. It is.
The light from the illumination device (2) is blocked by the transport drum (9) at portions other than the grooves and slits for holding the tobacco filter (1) of the transport drum (9) for transport. Therefore, the line CCD camera (3) does not receive light (or transmitted light) when the cigarette filter (1) does not pass through. Therefore, in FIG. 7, for example, when the cigarette filter (1) has not passed, as in the sub-scan numbers 0 and 18, the image data is 0.
Here, the recording processing unit (5) records, for example, the sub-scan number of 1024 sub-scans, the image data, and the feature amount. When recording the 1024th and subsequent sub-scans, the oldest sub-scan is recorded. The scan number, the image data, and the feature amount may be overwritten and recorded by the new sub-scan number, the image data, and the feature amount.

Next, the range identification processing unit (6) identifies a sub-scanning number range suitable for inspection from the feature quantity stored in the recording processing unit (5).
For example, in FIG. 7, the sub-scan number 1 in FIG. 7 indicates that the cigarette filter (1) is still detected by the line CCD camera (3) from the feature amount recorded in the recording processing unit (5). It is detected that the cigarette filter (1) has entered the detection range of the line CCD camera (3) from the sub scanning number 2. Further, it is detected by sub-scan number 18 that the tobacco filter (1) has passed outside the detection range of the line CCD camera (3).
For example, the range specifying processing unit (6) refers to the recording processing unit (5) at regular intervals to check the updated sub-scan number, image data, and feature amount, and the feature amount becomes larger than 0 to 0. The sub-scan number at which the cigarette filter has started to pass is detected based on the sub-scan number. Further, the range specifying processing unit (6) detects the sub-scanning number at which the passage of the tobacco filter is completed with the sub-scanning number where the feature amount is 0 from the non-zero value. Further, the sub-scan number between the start of passage and the end of passage is detected as the sub-scan number during which the tobacco filter is passing. As described above, the range specification processing unit (6) can detect the sub-scan number while the cigarette filter is passing.
For example, in FIG. 7, it can be seen from the feature amount that the cigarette filter passes for sub-scan numbers 2 to 17. In addition, the range specifying processing unit (6) determines whether there is a tobacco filter for each sub-scanning direction of the image data stored in the recording processing unit (5) based on the feature amount stored in the recording processing unit (5). It is possible to determine whether or not.

Next, the range identification processing unit (6) uses a sub-scanning number suitable for inspection based on the feature amount stored in the recording processing unit (5) in the sub-scanning direction section while the cigarette filter is passing. Specify a range of
For example, the range specifying processing unit (6) may be arranged in the vicinity of the center excluding a predetermined range at both ends of the sub-scanning number with respect to the sub-scanning number range in which it is determined that the cigarette filter exists. A range of sub-scan numbers suitable for inspection may be specified as the range. For example, the fixed range at both ends of the sub-scan number is set to 1/4 each of the scans where it is determined that the cigarette filter is present. In this case, in FIG. 7, the sub-scan number in which the cigarette filter exists is 16 sub-scans from 2 to 17, and the predetermined range at both ends is 1/4 of 16 sub-scans and 4 sub-scans. . Therefore, by subtracting 4 sub-scans from both ends of the sub-scan numbers 2 to 17, the sub-scan number range suitable for the inspection is specified as “6 to 13”.
Alternatively, for example, the range specification processing unit (6) may specify a range of sub-scanning numbers suitable for inspection as a range in which the feature amount is a predetermined value or more. For example, it is determined in advance as a range in which the feature amount is 80000 or more. In this case, the sub-scan number range suitable for the inspection is specified as “7 to 11” from the feature amount of FIG.
Alternatively, for example, the range specifying processing unit (6) may specify a range of sub-scanning numbers suitable for inspection as a constant range before and after the sub-scanning number having the maximum feature value. For example, the sub-scan number 9 has the maximum feature amount from the feature amount of FIG. When the predetermined range before and after the predetermined sub-scan number is 3, the range of sub-scan numbers suitable for inspection is specified as “6 to 12”.
Alternatively, for example, the range specifying processing unit (6) may be a range having a feature amount equal to or more than a predetermined ratio with respect to the maximum feature amount having the maximum feature amount. For example, the maximum feature amount at the time of sub-scanning number 9 at which the feature amount is maximum is 89424 from the feature amount in FIG. When the predetermined constant ratio is 80%, the value of 80% of the maximum feature amount is 71539.2. Therefore, in this case, the sub-scan number range suitable for the inspection is specified as “7 to 12” from the feature amount of FIG.
Next, the image extraction processing unit (7) extracts the image data from the recording processing unit (5) based on the sub-scan number range suitable for the inspection specified by the range specification processing unit (6), and extracts the extracted image. By adding the data in the sub-scanning direction, the image data used for measurement is synthesized, and the synthesized image data is output to the measurement processing unit (8).

  The measurement processing unit (8) inspects the tobacco filter (1) based on the combined image data output from the image extraction processing unit (7) and used for measurement. For example, the position of the boundary between a plurality of members constituting the cigarette filter (1) is detected, the length of the member that is the length between the boundaries is detected, and the detected position of each boundary and the length of the member are The cigarette filter (1) is inspected by comparing it with a predetermined reference value.

As described above, the range specifying processing unit (6) specifies the range of the sub-scan number suitable for the inspection from the feature amount, and the image extraction processing unit (7) is based on the range specified by the range specifying processing unit (6). By extracting the image data from the recording processing unit (5) and synthesizing the image data used for the measurement for adding the extracted image data in the sub-scanning direction, there is an effect of preventing the influence of disturbance such as noise in the image used for the measurement. is there.
In particular, noise may cause an unexpected peak in the feature amount due to disturbance. For example, when the filter (1) is transported on the transport drum (9), a gap is formed when the filter (1) is lifted from the transport drum (1), and light is leaked from the generated gap, and transmitted due to the leaked light. A peak (or saturation due to leaked light) may occur in image data for photographing light. In particular, in the image in the sub-scanning direction of the cigarette filter, peaks may occur at both ends of the sub-scan number range where it is determined that the cigarette filter exists. That is, noise such as a peak occurs in an image in the sub-scanning direction (time-axis image) during conveyance.
Even in such a case, for example, only the range near the center excluding a predetermined range at both ends of the sub-scan number with respect to the range of the sub-scan number where it is determined that a cigarette filter exists. By using, that is, by specifying the range in the time-axis image, there is an effect of preventing the influence of disturbance such as noise of the captured image.

Next, a detailed example as an example of detecting the position of the boundary of the members constituting the cigarette filter (1) based on the image data received from the range specifying processing unit (6), which is performed in the measurement processing unit (8). Processing will be described.
The measurement processing unit (8) first detects one end and the other end of the tobacco filter (1) in the image data. Here, one end is the position of the pixel corresponding to the end of the cylindrical light diffuser (1) in the image data, and is the pixel with the smaller pixel number in the image data (hereinafter referred to as the left end). On the other hand, the other end is the position of the pixel corresponding to the opposite end of the cylindrical light diffuser (1) in the image data, and is the pixel with the larger pixel number in the image data (hereinafter referred to as the right end). .
The left end is a pixel position as shown by edge1 in FIG. The outline of the detection method of the left edge edge1 can be detected by detecting the left edge portion where the pixel level is continuously low. This is because the end of the cylindrical light diffuser (1) is a dark member, the image level corresponding to the bright member is low, and the image level outside the cylindrical light diffuser (1) is high. .
First, the left end detection method will be described using the flowchart of FIG. 8 and image data as an example shown in FIG.
First, variables used in the flowchart of FIG. 8 will be described. x is the pixel position of interest currently being processed. Peak is a peak value of the light receiving level from the search start pixel to the target pixel. The slice is a slice level indicating a threshold value. RATIO is a predetermined constant ratio with respect to the peak used for calculating the slice level slice. cont is a count value of the number of pixels below the slice level, and is a variable for counting the number of consecutive pixels below the slice level slice. CONT is a threshold value that is a predetermined value for detecting the end by comparing with the number of consecutive pixels cont below the slice level. CONT has, for example, the number of pixels having the width shown as the width determination in FIG. 9B, and has a value suitable for detecting the end. PIXEL is the number of pixels of the synthesized image data, and is determined by the number of pixels of the line CCD camera (3). edge1 is the detected leftmost position of the tobacco filter (1) in the image data. Further, level (x) is a function for extracting the level of gradation of the pixel data at the position of the pixel x from the synthesized image data.
Further, the image data is increased by 1 in order of pixels from left to right, the pixels are identified by pixel numbers, the pixel number on the left end of the image data is 1, and the pixel number on the right end of the image data is PIXEL. explain.

Next, the operation for detecting the left end of the synthesized image data as an example shown in FIG. 9 will be described with reference to the flowchart shown in FIG.
First, the measurement processing unit (8) initializes the peak value peak, slice level slice, target pixel position x, and count value cont (step S801). The peak value peak is initialized as the level value level (1) of the pixel signal at the position of the pixel 1, the slice level slice is initialized as the peak value peak × ratio RATIO, the target pixel position x is 2, and the count value cont is 0. Is initialized as
Next, the measurement processing unit (8) determines whether or not the level value level (x) at the target pixel position x is larger than the peak value peak (step S802). If it is determined in step S802 that the level value level (x) at the target pixel position x is larger than the peak value peak, the peak value peak is set to the level value level (x) at the target pixel position x, and the slice level slice is set. Is the peak value peak × ratio RATIO (step S803). Next, the level value level (x) at the target pixel position x is compared with the slice level slice (step S804). If it is determined in step S802 that the level value level (x) at the target pixel position x is larger than the peak value peak, the level value level (x) at the target pixel position x is compared with the slice level slice. (Step S804)
Next, when the level value level (x) of the target pixel position x is smaller than the slice level slice in the determination in step 804, the measurement processing unit (8) increases the count value cont by one (step S805). Next, the count value cont is compared with the threshold value CONT (step S806). If it is determined in step 806 that the count value cont is smaller than the threshold value CONT, the target pixel position x is increased by one (step S807). If it is determined in step 804 that the level value level (x) of the target pixel position x is not smaller than the slice level slice, the count value cont is set to 0 (step S809), and then the target pixel position x is set. It increases by one (step S807).
Next, the measurement processing unit (8) compares the target pixel position x with the number of pixels PIXEL (step S808). If the target pixel position x is equal to or smaller than the pixel number PIXEL in the comparison in step S808, the processing from step 802 is repeated. If the target pixel position x is not less than or equal to the pixel number PIXEL in the comparison in step S808, the process ends as abnormal termination.
If it is determined in step 806 that the count value cont is not smaller than the threshold value CONT, the left edge position edge1 is calculated by the following formula (1) from the target pixel position x and the threshold value CONT (step S810).
edge1 = x-CONT (1)

As described above, the measurement processing unit (8) has a peak that is the level of the gradation level of the maximum image data among the image data from the small pixel to the target pixel position x in the direction from the small pixel to the large pixel. The value peak is searched, and the slice level slice is calculated by multiplying the searched peak value peak by a predetermined ratio RATIO.
Next, the measurement processing unit (8) detects pixels having a gradation level smaller than the calculated slice level slice, and the number of consecutive pixels having a gradation level smaller than the detected slice level slice is determined. A position at which the level of gradation of the pixel starts to become smaller than the slice level slice when it becomes larger than the predetermined threshold value CONT is detected as the left edge position edge1.

FIG. 9 shows a peak value peak and a slice level slice according to the flowchart described with reference to FIG. 8 for image data that has been combined and filtered as an example. FIG. 9A is an overall view of the image data, and FIG. 9B is an enlarged view of a portion near the left end extracted from the overall view of the image data. FIG. 9B also shows a peak value peak and a slice level slice.
For example, the threshold value CONT has a width indicated as a determination width. The slice level slice is determined when pixels whose gradation level is lower than the calculated slice level slice are continuously larger than a predetermined threshold value CONT (in this case, the same as the determination width in FIG. 9B). The left edge position edge1 is detected from the pixel level that has started to become smaller.

  Next, the measurement processing unit (8) detects the right end of the tobacco filter (1). The right end detection method is shown in the flowchart of FIG. When the measurement processing unit (8) detects the left end, the processing is performed from the small pixel toward the large pixel. However, when the right end is detected, the measurement processing unit (8) proceeds from the large pixel toward the small pixel. The difference is in the processing. Therefore, in step S801 in FIG. 8, the peak value peak is initialized as the level value level (1) of the pixel signal at the position of the pixel 1, and the target pixel position x is initialized as 2. In step S1001, the peak value peak is initialized as the level value level (PIXEL) of the pixel signal at the position of the pixel PIXEL, and the pixel position x of interest is initialized as PIXEL-1. In addition, in step 807 in FIG. 8, the target pixel position x is increased by one, whereas in step 1007 in FIG. 10, the target pixel position x is decreased by one. Further, in step S808 in FIG. 8, the target pixel position x is compared with the number of pixels PIXEL, whereas in step S1008 in FIG. 10, a comparison is made between the target pixel position x and the rightmost position of the image data. Is different. Further, in step S810 of FIG. 8, the left edge position edge1 is calculated from the target pixel position x and the threshold value CONT by edge1 = x−CONT, whereas in step S1010 of FIG. 10, the right edge position edge2 is calculated. The difference is that the calculation is performed with edge2 = x + CONT from the target pixel position x and the threshold value CONT.

As described above, the measurement processing unit (8) has the gradation level of image data that is the largest among the image data from the large pixel to the target pixel position x in the direction from the large pixel to the small pixel. The peak level peak is searched, and the slice level slice is calculated by multiplying the searched peak value peak by a predetermined ratio RATIO.
Next, the measurement processing unit (8) detects the level of gradation of pixels smaller than the calculated slice level slice, and the number of pixels having continuous gradation levels of pixels smaller than the detected slice level slice is determined. A position at which the level of the gradation of the pixel starts to become smaller than the slice level slice when it becomes larger than the predetermined threshold value CONT is detected as the right edge position edge2.

  The measurement processing unit (8) detects the left end position and the right end position by the above processing. Next, the measurement processing unit (8) specifies the measurement range for measuring the cigarette filter (1) as between the detected left end position and right end position.

Next, a measurement process part (8) detects by the image data which combined the boundary position of the member with which the transmittance | permeability from which the tobacco filter (1) differs, and the specified measurement range.
First, the measurement processing unit (8) applies a low-pass filter to the specified measurement range in the combined image data. As the low-pass filter, a filter that is shorter than the width of the change in the image data due to the length of the member and longer than the width of the peak of the image data due to the gap generated between the members or the width of the noise in the image data is used.
By applying a low-pass filter, it is possible to reduce a noise component included in the image data and stably detect a member boundary that gradually changes.
Further, by limiting the application range of the low-pass filter to the specified measurement range, the waveform of unstable image data due to transmitted light that has passed outside the tobacco filter (1) is excluded from the low-pass filter target, The transmitted light that has passed through the outside of the tobacco filter (1) can be suppressed from affecting the low-pass filter processing.
For example, when a low pass filter is applied to the image data shown in FIG. 11A, a signal as shown in FIG. 11B is obtained.

Next, the measurement processing unit (8) detects the maximum value MAX and the minimum value MIN from the image data to which the low-pass filter is applied. The maximum value MAX can be regarded as the light reception level of the transmitted light of a relatively bright member (bright member) among the members having different transmittances constituting the tobacco filter (1). Further, the minimum value MIN can be regarded as the light reception level of transmitted light of a relatively dark member (dark member) among the members having different transmittances constituting the tobacco filter (1).
Next, the measurement processing unit (8) calculates a member slice level SL that is a threshold for obtaining a member boundary position from the maximum value MAX and the minimum value MIN. For example, the member slice level SL may be an average of the minimum value MAX and the minimum value MIN. Alternatively, for example, the member slice level SL may be obtained by the following formula (2), where C is a constant indicating a predetermined value from 0 to 1.
SL = (MAX−MIN) × C + MIN (2)

Next, a measurement process part (8) searches each boundary position using member slice level SL calculated | required above. In the case of the tobacco filter (1) in the present embodiment, for example, it is known that there are four boundaries in advance as shown in FIG. In the image data, since the waveform of the dark member is more stable than the waveform of the bright member, the detection of the boundary changes from the dark member to the bright member, and detects a position exceeding the member slice level.
Therefore, in the detection performed by the measurement processing unit (8), the position (M) of the center portion of the specified measurement range in the measurement range specified as the position between the detected left end position (edge1) and right end position (edge2). Is a dark member. The position (M) of the central part is calculated by the following equation (3).
M = (edge1 + edge2) / 2 (3)
In the detection performed by the measurement processing unit (8), as shown in FIG. 11C, first, the boundary is detected in order from the center position (M) of the specified measurement range to the left end (D1), and the image is detected. A position where the data exceeds the member slice level SL is detected as a boundary ED1. Next, the boundary is detected in order from the center of the specified measurement range toward the right end (D2), and a position where the image data exceeds the member slice level SL is detected as the boundary ED2. Next, the boundary is detected in order from the left end to the right end (D3), and a position where the image data exceeds the member slice level SL is detected as the boundary ED3. Next, the boundary is detected in order from the right end to the left end (D4), and the position where the image data exceeds the member slice level SL is detected as the boundary ED4. Here, D1 to D4 in FIG. 11C indicate search directions.
Here, from the center (M) to the left end (D1), from the center (M) to the right end (D2), from the right end (M) to the left end (D3), from the left end to the right end (D4), The range in which each detection is performed is a detection range LS represented by the following expression (4) from each start point (center, right end, left end).
LS = (LR + LD) / 2 (4)
Here, LS is the length of the detection range, LR is the reference length of the bright member, and LD is the reference length of the dark member that is not the end member.

Next, the measurement processing unit (8) inspects the position of the boundary, the number of boundaries, and the length of each member between the boundaries from the detected boundary ED1 to ED4.
Moreover, a measurement process part (8) determines with the arrangement | sequence of a member being abnormal, when there is no pixel exceeding a slice level in the specified measurement range.
Moreover, a measurement process part (8) determines with the arrangement | sequence of a member being abnormal, when the value of the image data used as the center of the specified measurement range exceeds the member slice level SL.

As described above, according to the present invention, from a plurality of image data in the sub-scanning direction, a feature amount indicating the quality of image data is extracted for each sub-scan, and the image data has a feature amount effective for inspection. And adding the selected image data in the sub-scanning direction, even if noise is mixed into a small number of sub-scanning image data due to some influence, other sub-scanning image data that does not contain noise is also added. Therefore, normal inspection can be performed.
Further, by selecting and adding a plurality of gradation levels of image data in the sub-scanning direction in the sub-scanning direction, fluctuations in signals to be inspected for each inspection object are reduced, and a plurality of cylindrical light diffusers are Inspection can be performed with image data of the same part in the sub-scanning direction.
In addition, for each of the plurality of cylindrical light diffusers to be inspected, since the image data in the sub-scanning direction is from the same part on average, the appropriate threshold is the same for each cylindrical light diffuser, It is possible to always maintain an appropriate threshold value. It is also possible to compare image data of a plurality of cylindrical light diffusers.

For example, in the embodiment, the boundary position of the member is obtained by obtaining the member slice level SL by the following equation, where C is a constant indicating a predetermined value from 0 to 1.
SL = (MAX−MIN) × C + MIN
The value of C can be obtained statistically from image data obtained by photographing the cigarette filter with transmitted light and the actual boundary position. However, since the image data is unstable from the unstable image data every time the conventional measurement is performed, the values of MAX and MIN are also unstable, and therefore it is difficult to stably determine the value of C. . According to the present invention, it is possible to acquire stable image data, the values of MAX and MIN are also stable, and therefore C can be determined for a stable image.
Thus, by stabilizing the image data at each measurement, it is possible to compare image data in a plurality of measurements and perform statistical processing on the image data.

Next, the effect of the inspection method according to the present invention will be described by comparing the inspection results of the inspection method according to the present invention and the conventional inspection method.
In comparison, the photographing device (illumination device, CCD camera, image processing apparatus) and the photographing object are the same, and the image to be inspected is also the same.

The inspection object is a cigarette filter, the total length is 100 mm (low transmittance member 20 mm, high transmittance member 30 mm), and the conveyance speed is 3000 / min.
Here, the cigarette filter to be inspected is attached to the conveyance drum, the conveyance drum is rotated at a constant speed, and an image is taken using the above device, and the comparison is performed.

  In addition, the inspection method was confirmed under the following measurement conditions. The imaging conditions of the CCD camera are a drive frequency of 20 MHz, a scanning cycle of 0.5 msec, and a pixel resolution of 0.089 mm.

The condition for generating the image composition according to the present invention is that the feature amount is obtained from the area of the pixel data that is equal to or higher than the slice level, and the image composition range is 1/2 scan of the tobacco filter detection range (excluding 1/4 scan before and after). It is synthesized.
On the other hand, in a method using only an image in which a feature amount has a peak value (maximum value) according to a conventional method, only one-scan image data having a peak feature amount is obtained.

In the edge detection condition, the ratio (RATIO) is 0.5 (50%), and the count value (CONT) is 22 pixels (corresponding to 2 mm).
In the boundary detection condition, the ratio (C) is 0.5 (50%).

Under the measurement and detection conditions as described above, a cigarette filter as shown in FIG. 14A is placed on the transport drum and moved from the top to the bottom in the sub-scanning direction, and the transmitted light imaged using the CCD camera is captured. Images are shown in FIGS. 14 (b) and 14 (c).
Here, FIG. 14B shows a case where the cigarette filter is in close contact with the transport drum, and only light transmitted through the cigarette filter is imaged.
FIG. 14C is an image in the case where a gap is generated between the cigarette filter and the transport drum. In the cigarette filter image, the upper part (indicated by ca in FIG. 14C) is the image from the transport drum. Bright reflected light is photographed.
In the following description, an image taken in close contact as shown in FIG. 14B will be described as a close contact image, and an image obtained when light (reflected light) is taken in the image as shown in FIG. To do.

FIG. 15 illustrates feature amounts (vertical axis) calculated for each position (horizontal axis) in the sub-scanning direction, which is the traveling direction of the transport drum, with respect to the contact image in FIG. 14B and the gap image in FIG. ).
In FIG. 15, the feature amount is displayed on the graph only for the position in the sub-scanning direction where the cigarette filter is detected by the transmitted light. When the cigarette filter is conveyed at 3000 / min, the number of scans in the sub-scanning direction detected as having the cigarette filter is about 20.
From FIG. 15, the close contact image has a gentle change having a peak at the 9th sub-scan, whereas the gap image has a distorted change having a peak at the 4th sub-scan. This is because a gap occurs around the fourth sub-scan and the transmitted light is directly photographed.

FIG. 16 is a graph obtained by plotting a scanning image (peak image) with a feature amount peaking according to the prior art and a composite image according to the present invention from the contact image of FIG. 14 (b). In FIG. 16, the horizontal axis is the pixel position in the main scanning direction, and the vertical axis is the pixel level value of each of the peak image and the composite image at the pixel position in the main scanning direction.
Here, the peak image is image data of only the ninth sub-scan that has the maximum feature amount than the feature amount in the contact image of FIG.
In the composite image (composite image) according to the present invention, the scanning position in the sub-scanning direction is from the 6th scan to the 15th scan as 1/2 scan (excluding front and rear 1/4 scan) of the tobacco filter detection range. This is a pixel level calculated from image data for 10 scans (in this case, the calculation is an average).

In FIG. 16, for example, when Th1 is set as the threshold value, the pixel positions in the main scanning direction of the boundary positions Z1, Z2, Z3, and Z4 are detected from the intersections of the pixel level waveform of the peak image or the synthesized image and Th1. It becomes possible.
As described above, in the contact image, it is possible to detect the boundary position in the main scanning direction in both the peak image and the composite image.

  However, it can be seen from FIG. 16 that an image with a reduced noise component can be obtained by using a composite image as compared with the peak image, although the image level is somewhat lowered. It can also be seen that a stable image can be obtained even in the repeated test. These are particularly noticeable when the pixel position in the main scanning direction is the position before and after the center of 300 and the position before and after the center of 900.

Next, FIG. 17 is a graph in which a peak image and a composite image are plotted from the gap image in FIG. In FIG. 17, the horizontal axis represents the pixel position in the main scanning direction, and the vertical axis represents the pixel level value at the pixel position in the main scanning direction of each of the peak image and the composite image.
Here, the peak image is image data only in the fourth sub-scan where the feature amount is maximum compared to the feature amount in the gap image in FIG.
Further, in the composite image (composite image) according to the present invention, the image data for 10 scans from the 6th scan to the 15th scan is used as 1/2 scan (excluding front and rear 1/4 scan) of the tobacco filter detection range. This is the calculated pixel level.

In FIG. 17, as in FIG. 16, for example, when Th <b> 1 is set as the threshold, the main scanning directions of Z <b> 1, Z <b> 2, Z <b> 3, and Z <b> 4 that are boundary positions from the intersection of the pixel level waveform of the peak image or the synthesized image and Th <b> 1. Can be detected in a composite image.
However, the position cannot be normally detected in the peak image.

That is, from FIG. 17, in the method of measuring from an image having a peak feature amount (conventional method), when a gap occurs, measurement processing is performed from an image in which the cigarette filter is not properly imaged. No result can be obtained (position cannot be detected).
On the other hand, by using the composite image according to the present invention, even when the light leaks due to, for example, a gap as shown in FIG. A measurement image can be obtained (position can be detected).

  As described above, by using the detection method according to the present invention (using a composite image), for example, when a gap occurs and noise (such as light leakage) occurs, the boundary position of the cigarette filter is normally detected. As a result, the cigarette filter can be normally inspected.

  In the description of FIGS. 16 and 17, the threshold value is used and the boundary position is detected from the intersection with the threshold value. However, the detection position is detected using the boundary position detection method according to the method of the present invention. Also good.

  As a result of repeatedly measuring the same cigarette filter, the detection reproducibility of the edge and each boundary position was 1 pixel or less (0.089 mm or less) in the main scanning direction. Further, by stabilizing the image by image synthesis and detecting the edge and boundary position relative to the image level, for example, the image level of the high transmittance member (bright member) is in the range of about 50 to 200 If it is within the range, stable measurement processing is possible (the lower limit and the upper limit of the image level are set to 0 and 255).

As described above, according to the inspection method of the present invention, it is possible to obtain a stable measurement image from an image signal that constantly fluctuates, and the measurement position accuracy can be improved.
In addition, in the conventional measurement with a fixed slice level, it is necessary to keep the image level at an appropriate position, and frequent condition adjustments are necessary as the transmittance changes due to material changes and the illumination brightness decreases due to dirt and deterioration. It was. By using the inspection method according to the present invention, even if the image level varies due to material changes or light quantity variations, the measurement process is performed according to the image level, so that the maintenance frequency can be greatly reduced.

  In the description of the present invention, a cylindrical light diffuser has been described. However, the present invention is not limited to a cylindrical light diffuser and can be used for inspection of a light diffuser of any shape. It is. In addition, the inspection method according to the present invention is an object to be inspected in which members having different transmittances are alternately arranged, and can be applied when the lengths of a plurality of alternately arranged members are known in advance. is there.

  In the description of the present invention, the range specifying processing unit (6) specifies the range of sub-scanning numbers suitable for inspection from the feature values stored in the recording processing unit (5). (6) may detect only the sub-scan number having the maximum feature value, and the measurement processing unit (8) may perform subsequent processing using only the image data of the sub-scan number having the maximum feature value. .

The inspection apparatus 10 may be realized by dedicated hardware, and the inspection apparatus 10 includes a memory and a CPU (central processing unit) to realize the function of the inspection apparatus 10. This function may be realized by loading the program into a memory and executing it.
The recording processing unit 5 may be configured by a nonvolatile memory such as a hard disk device, a magneto-optical disk device, or a flash memory, a volatile memory such as a RAM (Random Access Memory), or a combination thereof. .

  The embodiment of the present invention has been described in detail with reference to the drawings. However, the specific configuration is not limited to this embodiment, and includes design and the like within a scope not departing from the gist of the present invention.

  The present invention is suitable for use in a tobacco filter inspection apparatus.

It is a block diagram which shows the structure of a cylindrical light-diffusion body. It is a block diagram which shows the structure of the test | inspection apparatus by one Embodiment of this invention. It is explanatory drawing which shows the waveform of the transmitted light of a cylindrical light diffuser. It is a block diagram which shows the structure for demonstrating operation | movement of the test | inspection apparatus by one Embodiment of this invention. It is explanatory drawing which shows the 1st calculation method of the feature-value. It is explanatory drawing which shows the 2nd calculation method of the feature-value. It is a content figure which shows the content of the data recorded on a recording process part. It is a flowchart figure which shows the flow of the process which detects a left end. It is explanatory drawing explaining the process which detects a left end. It is a flowchart figure which shows the flow of the process which detects a right end. It is explanatory drawing explaining the method to detect a boundary. It is a block diagram which shows the structure of the conventional inspection apparatus. It is explanatory drawing explaining the detection method by the conventional test | inspection apparatus. It is explanatory drawing explaining the measurement image of a tobacco filter. It is a graph which shows the change of the feature-value in the case of a contact | adherence image and a clearance gap image about FIG. 6 is a graph comparing pixel levels according to the present invention with those in a contact image. It is a graph which compares the pixel level by a conventional and the present invention in a crevice picture.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Cylindrical light diffuser, tobacco filter 2 Illuminating device 3 Line sensor, line CCD camera 4 Feature extraction circuit 5 Recording circuit 6 Measurement processing 7 Waveform example 8 Cigarette filter 9 Conveyance drum 10 Inspection device 32a Slit 33 Light guide 34 Light source 36 Dimming member

Claims (4)

An inspection device for a light diffuser that conveys a longitudinal light diffuser in a sub-scanning direction perpendicular to the longitudinal direction and inspects the light diffuser in a main scanning direction parallel to the longitudinal direction,
Light source means for irradiating the light diffuser with light in a scanning direction;
Each time the light diffuser is transported in the sub-scanning direction, the light diffuser is arranged so as to receive light emitted from the light source means and is transmitted from the light source means and transmitted through the light diffuser. Photographing means for photographing and outputting as image data in which pixels are arranged in a line;
A pixel having a gradation level of image data in the sub-scanning direction outputted from the photographing means is detected in the main scanning direction from the image data, and is characterized based on the detected pixel in the main scanning direction. Feature amount extraction means for calculating the amount for each image data;
A range specifying unit for extracting a range of image data in the sub-scanning direction from the image data in the sub-scanning direction based on the feature amount calculated by the feature amount extracting unit;
Image extraction means for adding the image data in the sub-scanning direction based on the range of image data in the sub-scanning direction extracted by the range specifying means, and generating image data for inspection;
An inspection apparatus comprising: It will shift the target pixel from one end of the image data for inspection by the image extracting unit has generated, with determining the peak level which is the maximum value of the gradient to the target pixel from one end, multiplied by a constant ratio to the peak level a first detecting means for detecting a position with a value of threshold, corresponding to pixel positions of pixels less than the threshold value are continuous to one end of the previous SL light diffuser,
The pixel of interest is shifted from the other end of the image data for inspection generated by the image extraction means to obtain a peak level that is the maximum value of the gradation from the other end to the pixel of interest, and a constant ratio is set to the peak level. a second detecting means for detecting a value obtained by multiplying the threshold value, as a position corresponding to the other end of the prior SL light diffuser pixel position where a pixel to be smaller than the threshold value are continuous,
From the range between the one end position detected by the first detection means and the position of the other end detected by the second detection means, the main scanning direction corresponding to the light diffuser portion of the image data for inspection Main scanning direction range specifying means for specifying a main scanning direction range that is a range;
The inspection apparatus according to claim 1, further comprising: Boundary image data extracting means for extracting image data for inspection generated by the image extracting means based on the main scanning direction range specified by the main scanning direction range specifying means;
Boundary threshold value calculating means for obtaining a maximum value and a minimum value of the image data extracted by the boundary image data extracting means, and calculating a threshold value by multiplying a difference between the obtained maximum value and minimum value by a certain ratio;
Boundary position specifying means for specifying the boundary position of the light diffuser member based on the image data extracted by the boundary image data extracting means and the threshold value calculated by the boundary threshold value calculating means;
The inspection apparatus according to claim 2, further comprising: It is a method used in an inspection device for a light diffuser that transports a longitudinal light diffuser in a sub-scanning direction perpendicular to the longitudinal direction and inspects the light diffuser in a main scanning direction parallel to the longitudinal direction,
A light source means irradiates the light diffuser with light in a scanning direction;
An imaging unit is arranged to receive the light emitted from the light source unit, and the light diffuser is transported in the sub-scanning direction by the transmitted light which is the light emitted from the light source unit and transmitted through the light diffuser. Each time it is taken, and output as image data with pixels arranged in a line,
A feature amount extraction unit detects pixels having a gradation level of image data in the sub-scanning direction output from the imaging unit in a main scanning direction from the image data, and the detected main scanning direction is detected from the image data. A feature amount is calculated for each image data based on the pixels of
A range specifying unit extracts a range of image data in the sub-scanning direction from the image data in the sub-scanning direction based on the feature amount calculated by the feature amount extracting unit,
An image extraction unit adds the image data in the sub-scanning direction based on the range of image data in the sub-scanning direction extracted by the range specifying unit to generate image data for inspection.
Inspection method characterized by that.

Images