JP2002074332A - Method and device for detecting sealing defect of packing material and packing body - Google Patents

Abstract

PROBLEM TO BE SOLVED: To precisely discriminate each sealing defect of a packing bag having the sealing part of an opaque packing bag by adopting a variable density picture processing means and another processing method. SOLUTION: Relating to the sealing defect detector, a seal part is photographed by an illuminator and a camera, this photographic data are subjected to feature extracting processing by variable density processing, or feature extraction processing by fuzzy picture processing, or feature extraction processing by morphological filter or feature extraction processing by fuzzy morphological filter using an adaptive structure element and sealing defects are discriminated by pattern matching processing by using these respective processings.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a wrapping material having a sealing portion, a wrapping material for detecting and discriminating a sealing failure such as biting, wrinkling, misalignment, water bubbles, air bubbles and the like, and a pinhole.
The present invention relates to a method and an apparatus for detecting a defective seal of a package.

[0002]

2. Description of the Related Art The prior art comprises a step of continuously transporting a packaged product having a heat-sealed portion at an end portion, and a step of transferring the packaged material and the heat-sealed portion of the package through a predetermined position during the transport. Imaging using a light beam transmitted through the heat seal portion, and performing A / D conversion on an image signal obtained by the imaging, as multi-level gradation data corresponding to the transmitted dose over the entire surface of the heat seal portion. Storing, comparing the stored value stored as the gradation data with a threshold value set according to a separate condition, and setting the stored value equal to or less than the threshold value to 0,
Storing the above as 1/1 0/1 data, and outputting a failure signal when the 0/1 data does not pass a predetermined defective heat seal portion determination criterion for the entire surface of the heat seal portion. There is a method of selecting packaging materials and packages having a heat seal portion by binary image processing.

[0003]

However, in the above-mentioned prior art, a light beam transmitted through the heat seal portion is applied.
Since the determination is made using the data of the intensity of the transmitted light, there is a problem that it is difficult to select an opaque wrapping material which does not transmit the light and has a large amount of irregularly reflected light, or a defective seal portion of the wrapping body.

[0004] In view of such circumstances, the present invention provides a feature extraction process by shading image processing, which detects biting, wrinkles, misalignment, poor sealing such as water bubbles and air bubbles, and pinholes in a sealing portion.
Morphological filter (Mathematical)
Feature extraction processing by Morphological Filters, feature extraction processing by fuzzy image processing, fuzzy morphological filter (Fuzzy)
Mathematical Morphologica
l Filters), a packaging material, a method for detecting a seal failure of a package, and a seal failure detection device for performing a feature extraction process using a fuzzy morphological filter using an adaptive structure element and discriminating the feature extraction using pattern matching. The purpose is to provide.

[0005]

SUMMARY OF THE INVENTION The present invention has been made to solve the above problems, and employs the following technical means. According to the first aspect of the present invention, a technical means is employed in which a seal failure and a pinhole, in which it is difficult to determine the seal portion, are determined by pattern matching using feature extraction by grayscale image processing.

According to the second aspect of the present invention, a seal failure and a pinhole in which a seal portion is difficult to determine are removed by using a morphological filter (Material Morp).
Technical means for discriminating using pattern matching using feature extraction by holographic filters is employed.

According to the third aspect of the present invention, a technical means is employed in which a seal failure and a pinhole in which it is difficult to determine a seal portion are determined by using pattern matching using feature extraction by fuzzy image processing.

According to the fourth aspect of the present invention, a seal failure and a pinhole in which a seal portion is difficult to determine are eliminated by a fuzzy morphological filter (Fuzzy Mathematical Filter).
Tical Mathematical Morpho
A technical measure of discriminating using pattern matching using feature extraction by logical filters is employed.

According to the fifth aspect of the present invention, a seal failure and a pinhole in which it is difficult to determine a seal portion are determined by pattern matching using feature extraction by a fuzzy morphological filter using an adaptive structural element. The technical means called was adopted.

In the invention according to the sixth aspect, a technique is adopted in which, for a seal failure and a pinhole in which it is difficult to determine the seal portion, the seal portion is photographed by using illumination that is illuminated obliquely, and features are extracted and determined. Means were adopted.

In the invention according to claim 7, in the defect detecting device for the sealing portion of the packaging material and the package manufactured by the packaging machine, the bag making machine or the like for performing the sealing, the sealing portion is irradiated obliquely. Technical means were adopted.

According to the present invention, in a defect extracting device for a sealing portion of a packaging material and a package manufactured by a packaging machine, a bag making machine or the like for performing sealing, a lighting device for irradiating the sealing portion obliquely. Is installed, a camera for photographing the seal part is installed, the seal part is photographed by the lighting device and the camera, and the photographed data is subjected to feature extraction processing by grayscale image processing, feature extraction processing by fuzzy image processing, or morphological. Perform a feature extraction process using a filter, a feature extraction process using a fuzzy morphological filter, or a feature extraction process using a fuzzy morphological filter using an adaptive structuring element, and discriminate by pattern matching using each processed image. The technical means called was adopted.

[0013]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The detection of defective seals, particularly the detection of defective seals of opaque packaging materials, is currently mostly performed by visual inspection.
There is a case where a transmitted light is output to the seal part of the transparent bag and the judgment is made by using the transmitted data.However, a method and an apparatus for judging by using pattern extraction using feature extraction by image processing are performed. Not. For example, poor sealing of an opaque retort package or the like is difficult to discriminate because there are bites, wrinkles, gaps, water bubbles, air bubbles, and the like. And packaging materials,
Since there is an aluminum layer in the material of the package, noise for detection such as uneven illumination and sagging of the bag is likely to be mixed. There is a need for a seal failure detection method and a seal failure detection device for removing these and detecting defects.

In addition, the conventional method of detecting a defective seal cannot detect defects in many cases, resulting in poor inspection efficiency and the like. In addition, it is indispensable to guarantee the hygiene of the product and the function of the seal. In addition, 100% inspection is indispensable for speeding up the machine in the future, and automation of the inspection is desired.

Therefore, the following five algorithms have been developed as shown in FIG. 5 as a method for judging and inspecting the sealing failure of the packaging bag. (1) A defect discrimination method by grayscale image processing. (A) Feature extraction by density gradation conversion By increasing the contrast of an image, the image can be easily viewed and the defect can be easily identified. This is an operation of mapping a density value to another density value by a function or a conversion rule. (B) Feature extraction by sharpening For an image in which a defective edge is blurred, a method of removing the blur and further enhancing the edge is used.
The feature of the defect is emphasized. As the calculation, there is a Laplacian method or the like. (C) Feature extraction by smoothing Defects are emphasized by performing smoothing as a method of removing or reducing noise present in an inspection image. The operation includes a median filter, a selective local averaging method, and the like. (D) Feature Extraction by Filtering The density value group of the inspection image is processed by a certain filter, and a defect feature is extracted by an operation (filtering) that makes the processed result an output value. In the present invention, a defect is extracted by using a filter for edge detection, for example, a differential filter, a line detection filter, pattern matching, or the like,
A good product is determined. (2) Morphological filter (Mathemati)
cal Morphological Filter
s). (3) A defect determination method using fuzzy image processing. When the boundary between a defect and a region other than the defect is ambiguous in the inspection image, features are extracted by using a fuzzy image. The extraction method is essentially the same as the method using the grayscale image processing, but is characterized by treating the image as a fuzzy image. (4) Fuzzy morphological filter (Fuzzy)
Mathematical Morphological
al Filters) (hereinafter, referred to as FMF). (5) This is a discrimination method by FMF using an adaptive structure element (hereinafter, referred to as ASE). Each of these five algorithms is effective in the case as shown in FIG.

Next, an outline of the system will be described. Immediately after the heat seal, the extraction system of the present invention
The defect is imaged by a solid-state image sensor (CCD) and detected by image processing.

The characteristics are listed below, and the inspection object is effective for opaque bags (aluminum foil, vapor-deposited, paper, etc.) as packaging materials and packages. It is a method and an apparatus for detecting a seal failure which is not affected by the type of failure and the degree of accuracy required. In principle, the present invention can be applied to various packaging bags other than the opaque bags. The inspection position can be adapted to the inspection of the heat seal portion (upper, lower, side), and the inspection items include bite, incomplete seal, wrinkle, bag misalignment, poor seal width, water bubbles, air bubbles, and pins. Targets the hall.

Specifically, it is difficult to discriminate between a poor biting and a sagging of the bag, and it is difficult to discriminate a small bite. Further, the present invention has solved the problem that it was difficult to distinguish between wrinkles and bag deviation, wrinkles and illumination unevenness, illumination fringes, and the like.

The size of the bag and the width of the seal are variable by changing the optical system, and can be directly attached to an ordinary packaging machine.

Next, each constituent device will be described. The image processing apparatus performs processing by a personal computer, and uses a high-speed image board for image input / output and some image processing calculations. As shown in FIG. 6, the image pickup device 2 is inputted by using two general commercially available cameras (there is no need for a high-end camera that can capture a wide range with one camera). For example, the camera A is the left sticker, and the camera B is the right sticker. The part is imaged. Thereby, it is possible to inspect for finer defects.

With respect to the lighting device 1, the choice of the lighting method is very important in image processing, and the target packaging material cannot properly extract defective features due to its reflection under normal lighting. .

As an illuminating method for solving this, as shown in FIG. 7, when an oblique illuminating method of irradiating obliquely upward from below is used, the sealing defect is optically emphasized and discrimination becomes easy.
The illuminating light is inexpensive and exerts a tremendous effect on the inspection of the sealing failure of the packaging bag by using general fluorescent light and halogen illuminating light. By adopting the oblique illumination method of irradiating obliquely upward from the lower part, it is possible to clearly determine a seal defect that is difficult to determine, and to emphasize the defect.

Next, an example of the operation of the inspection system will be described with reference to FIG. When the inspection start button is pressed to start, for example, the rotating body of the rotary heat seal packaging machine is driven, the sealed packaging bag is detected by the sensor, and if the presence of the packaging bag is determined to be NO, the packaging bag is detected by the sensor. It is fed back to the process before the detection and detects again. If the presence of the packaging bag is YES, the image is captured by the electronic shutter of the camera, and then the failure is determined by image processing. If the presence of the failure is YES, the defective position is displayed on the display and the defective product is discharged by a signal. I do. Further, the defect information is output to a file. With this output, the number of defects on the detection screen is counted up, the number of inspections on the inspection screen is counted up, and if the total number of defects> set value is YES, the process ends.

If the defect is NO, the product is not discharged, and the process jumps to a step after counting up the number of defects on the inspection screen. If the total number of defects> set value is NO, the inspection is performed. It is designed to jump to the process after the start.

Next, the gradation image processing will be described with reference to FIG. Conventionally, a binarization method has been adopted as a defect detection method. This method can extract features by simple processing only for those with clear defects, but is less effective for minute defects.

Therefore, a grayscale image, that is, 64 gradations, 12
This is performed using a multi-tone image such as an 8-gradation, 256-gradation, 512-gradation, or 1024-gradation image. First, the process using a 256-tone image will be described with reference to FIG. 3. (Step 1) The input image sets an inspection area in each image with 256 × 256 pixels. This inspection area is 4 areas, area 1 is bag slippage and insufficient sealing, area 2
3 is biting / wrinkling, and the area 4 is insufficiently sealed, and four areas are provided by two cameras. Therefore, the inspection is determined in eight areas per bag (the number of areas is the inspection efficiency,
It can be increased or decreased depending on the inspection speed.)

First, a description will be given of an inspection method using grayscale image processing according to the first embodiment of the present invention. First, the rotating body in FIG. 1 is driven, a package is detected by a sensor, an image is captured by an electronic shutter, and inspection / discrimination is performed by grayscale image processing.

As a pre-process performed before the start of the inspection, an inspection area is set from the four areas described above. The reference template (example of a non-defective product differs for each inspection area). Set. This parameter determines how much failure is detected.

(Step 2) Since a reference template is different for each inspection area, a template is set for each area. The size of the template varies depending on the bag to be inspected.

(Step 3) Next, a determination threshold value for performing a good / bad determination based on the correlation value is set. This parameter determines how much failure is detected.

The above processing is performed as preprocessing before the start of inspection.

(Step 4) Next, a flow performed during the inspection will be described. First, an inspection is started on the operation screen (touch panel).

(Step 5) Then, as a gray-scale image capturing operation, the sticker portion is imaged by the two cameras by a synchronization signal from the sensor, and the left and right gray-scale images are captured. At this time, the position of the inspection image is aligned with the template such that the reference coordinates of the inspection image are the same as those of the template (position correction).

(Step 6) If necessary,
After performing pre-processing such as filtering to emphasize features, a normalized correlation process is performed in each region. This processing includes the following equation 1 between the template and the inspection window for each region.
Is calculated. Note that u and v are X and Y coordinates, respectively, N
Indicates the number of effective pixels of the template. In the present invention, the SSDA method (Sequential Simila) is used.
right Detection Algorithm)
Increases the calculation speed.

(Equation 1)

(Step 7) Then, the minimum value (maximum value) of the correlation value is obtained for each area.

(Step 8) Image A (left side) and Image B
The minimum value (maximum value) of the entire minimum value (maximum value) is obtained from the minimum value (maximum value) of each area (right side). That is,
The minimum value of the correlation value of the entire seal portion is obtained.

(Step 9) If the minimum value (maximum value) of the entire seal portion is smaller than the threshold value set in the preprocessing (Step 3), it is determined that a defect exists. That is, the correlation value (degree of coincidence with the template) is 1
If at least one is low, it is determined to be defective.

The minimum value and the maximum value of the correlation value are basically the same, but the detection rate may change depending on the conditions such as the difference between the packaging bags. Choose or be careful. What is important is that the criterion for determining good or bad changes depending on the threshold value.

(Step 10) If a defect exists, its position is displayed on the inspection screen.

(Step 11) The packaging bag number, the defective position (coordinate) on the image, and the correlation value are written in a file.
This can be displayed as a text file after the inspection.

(Step 12) Then, the total number of inspections and the number of defects are displayed on the inspection screen as inspection results.

Here, the features of the inspection principle will be described below.
Since the target defect is difficult to determine by binarization or the like, the defect feature is extracted from the grayscale image. A defect is determined by utilizing the fact that the correlation value of a defective portion is reduced by performing pattern matching with a template (a non-defective product) for each inspection area. A thing with a low correlation value in each area is likely to be a correlation value in a defective part. Therefore,
The minimum value of each area is obtained, and the minimum value is further calculated. Thereafter, good or bad is determined based on the magnitude relationship between the threshold value uniquely set for each inspection target and the obtained minimum value. The discrimination rate can be changed by setting the threshold.

Next, as a second embodiment, an inspection method using FMF will be described. FMF is an operation of a set called an object and a structuring element.
This is effective when only defects are extracted by effectively selecting structural elements, and four operations, Fuzzy
erosion, Fuzzy dilation, F
uzzy opening, fuzzy closin
g.

The definition formula (Formula 2) of FMF is shown below.

(Equation 2)

In the second embodiment, feature enhancement using fuzzy morphology processing with a fuzzy image was performed in order to enhance the features of minute defects. In addition to the preprocessing performed in the first embodiment, fuzzification parameters are set.

Here, the setting of the fuzzification parameter will be described. In this system, a grayscale image is converted into a fuzzy image using a standard π function. Each pixel of the image originally has a characteristic such as brightness, and the brightness is originally a fuzzy amount. The membership function assigns a fuzzy grade to an object. FIG. 8 shows the standard π function. Assuming that the point where the membership value is 0 in FIG. 8 is (m, n), the pixel X _mn is a fuzzy set A
Means that the point having a membership value of 1 belongs to the fuzzy set A.

If 0 <μ _mn <1, the pixel, that is, the pixel X _mn (μ _mn ) belongs to the fuzzy set A with the possibility μ _mn . In particular, μ _mn (X _mn )
When = 0.5, the pixel X _mn belongs to or is in the middle of not belonging to the fuzzy set A, and is the point at which the ambiguity is maximum, and is called a crossover point.

[0048] The determination of the cross-over is very important in fuzzy images, the degree of ambiguity in pixel X _{m n} to determine whether a member of A is because an intermediate. In this embodiment, the crossover point is determined from the histogram of the image including the target seal portion.

After setting the fuzzy parameters, the grayscale image is fetched, the grayscale image fetched into the memory is converted into a fuzzy image based on the setting of the fuzzy parameters, and feature enhancement is performed using the FMF. In particular, fuzzy opening and fuzzy closing, or a combination thereof are effective.

Then, a normalized correlation is performed between the template for each area and the inspection image on which the feature enhancement has been performed, and a defect is determined.

The calculation of the correlation value, the display of the defective position, the output of the file, the display of the number of inspections, and the like are the same as those in the first embodiment, and therefore, the detailed description is omitted.

Next, a description will be given of an inspection method using FMF by ASE according to the third embodiment. In the discrimination method using FMF, since the structural element is fixed, it is difficult to detect an ambiguous minute defect. Therefore, it is possible to flexibly respond to defects in which the shape or value of the structural element changes according to the target,
AS that can detect bending defects and minute defects
An inspection method using FMF by E is performed. Also, since the structural elements are automatically constructed, there is no need for preprocessing.

Next, the procedure of FMF processing by ASE will be described. Determine the membership values of the structuring elements using a standard scan from pixel to pixel. In addition,
The value of SE changes depending on the gray level of the image. ASE
Of membership value μ (x _ase) is defined as follows. μ _ase = 1− (μ (x _mn ) −μ (x _avr )) where μ (x _mn ) is the membership value of the target point, μ
(X _avr ) is the average of the membership values within 8 neighborhoods. With this, when the target pixel has a higher (lower) membership value than the average, the membership value of the structural element is allocated to a small value, and when the target pixel has a value similar to the average, the membership value is allocated to a large value. it can.

Next, the type of the structural element is determined. There are various forms of poor sealing, but in general, many are stick-shaped or circular. If the structural element matches the shape, fuzzy opening will give an emphasis on the defect, but the other parts will be removed. Therefore, 4 shown in FIG.
An operation is performed on each of the two structural elements, and the one having a high degree of coincidence is selected by taking the maximum value.

Next, a description will be given of a flow of defect determination by the FMF using the adaptive structure element with reference to FIG. In addition to the preprocessing performed in the first embodiment, fuzzification parameters are set. Then, image processing by FMF using ASE is performed. An enhanced image X _ASE obtained by the FMF processing using the _ASE is obtained by the following Expression 3.

(Equation 3)

When the feature emphasis is completed, the total number of pixels having a membership value equal to or greater than the set threshold value is obtained in the target window, and if it is larger than the set value, it is determined that a defect exists. This step can be performed by a normalized correlation process or the like.

Note that the display of the defective position, the output of the inspection result file, and the display of the number of inspections are the same as those in the first embodiment, and a detailed description thereof will be omitted.

Specialized knowledge is required for setting the conditions of this system. Therefore, in order for the on-site operator to operate the system effectively, a device such as an interactive setting method of conditions is required. Further, by adding a learning function such as a neural network or fuzzy function, automatic setting and creation of a user's own standard become possible.

Note that the inspection device of the present system does not necessarily need to be attached to a packaging machine by utilizing a network or the like, but the lighting / imaging device needs to be attached to the packaging machine. It is necessary to provide a space and mechanism for this on the packaging machine side.

[0060]

According to the present invention, as a result of employing the above configuration,
The following effects can be obtained. (1) By employing the shading image processing method, it is possible to accurately determine each sealing defect of a packaging bag having a sealing portion of an opaque packaging bag. (2) By adopting the feature extraction method by FMF, it is possible to accurately determine each seal failure of the packaging bag. (3) By adopting the feature extraction method by FMF using ASE, each sealing failure of the packaging bag can be determined with high accuracy. (4) By employing a feature extraction method using a morphological filter, it is possible to accurately determine each seal failure of a packaging bag. (5) Each seal failure of the packaging bag can be emphasized by the lower oblique illumination method. (6) By employing the fuzzy image processing method, it is possible to accurately determine each seal failure of the packaging bag.

[Brief description of the drawings]

FIG. 1 is a flowchart showing an operation flow of an inspection device of the present invention.

FIG. 2 is a flowchart showing a flow of defect determination by the grayscale image processing of the present invention.

FIG. 3 is a plan view showing an inspection area according to the present invention.

FIG. 4 is a flowchart showing a flow of defect determination by the FMF of the present invention.

FIG. 5 is a flowchart showing algorithm selection according to the present invention.

FIG. 6 is a plan view showing an arrangement relationship between an illumination device and an imaging device according to the present invention.

FIG. 7 is a front view illustrating an arrangement relationship between the illumination device and the imaging device according to the present invention.

FIG. 8 is a flowchart showing a flow of defect determination by FMF using the ASE of the present invention.

FIG. 9 is an explanatory diagram showing types of structural elements of the present invention.

[Explanation of symbols]

 1 illumination device 2 imaging device

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B057 AA02 BA19 CA08 CA12 CA16 CB08 CB12 CB16 CC01 CE03 CE05 CE06 DA03 DB02 DB09 DC05 DC22 DC34 DC36 5L096 AA06 BA03 CA05 EA01 FA00 GA55 GA57 HA08 JA09

Claims (8)

[Claims] 1. A method for detecting a seal failure of a packaging material and a package, wherein a seal failure and a pinhole in which a seal portion is difficult to determine are determined using pattern matching using feature extraction by grayscale image processing. . 2. A morphological filter (Mathematic filter) is used to detect a seal failure and a pinhole in which a seal portion is difficult to determine.
atomic MorphologicalFilter
A method for detecting a seal failure of a packaging material and a packaging body, characterized by using feature extraction by rs) and determining using pattern matching. 3. A method of detecting a seal failure of a packaging material and a package, wherein a seal failure and a pinhole in which a seal portion is difficult to determine are determined using pattern matching using feature extraction by fuzzy image processing. . 4. A fuzzy morphological filter (Fu morphological filter) for detecting a defective seal and a pinhole in which a seal portion is difficult to determine.
zzy Mathematical Morpholo
A method of detecting a seal failure of a packaging material and a packaging body, characterized by using pattern extraction and feature extraction by G.G. 5. A packaging material characterized in that a seal failure and a pinhole in which a seal portion is difficult to determine are determined using pattern matching using feature extraction by a fuzzy morphological filter using adaptive structural elements. And a method for detecting a seal failure of a package. 6. A method for detecting a defective seal, comprising: photographing a sealed part by using illumination for obliquely irradiating a defective seal and a pinhole in which it is difficult to determine the seal part; . 7. An illuminating device for illuminating a seal portion obliquely in a defect detecting device for a seal portion of a packaging material and a package manufactured by a packaging machine, a bag making machine, or the like for performing sealing. 8. A defect detecting device for a seal portion of a packaging material and a package manufactured by a packaging machine, a bag making machine, etc. for performing sealing, an illumination device for irradiating the seal portion obliquely is installed, A camera for photographing is installed, a seal portion is photographed by the lighting device and the camera, and the photographed data is subjected to feature extraction processing by grayscale image processing, feature extraction processing by fuzzy image processing, or feature extraction processing by morphological filters, or A sealing defect characterized by performing feature extraction processing using a fuzzy morphological filter or feature extraction processing using a fuzzy morphological filter using an adaptive structuring element, and using each processed image to determine by pattern matching processing. apparatus.