JPH04270951A - Method for inspecting bottle - Google Patents

Abstract

PURPOSE:To accurately detect the foreign matter present in the bottom part of an empty bottle or a liquid filled bottle by removing a scratch noise from the first image of the inner and outer surfaces of the bottom part of the bottle and the foreign matter in the bottle before comparing the first image with the second image only of the outer surface of the bottom part of the bottle. CONSTITUTION:A part of the light from a first light source 2 passes through the bottom part of a bottle 1 to be inspected while it is refracted and reflected in the body part and interior of the bottle 1 to be inspected and the knurling 10 of the bottom part of the bottle and the foreign matter in the bottle are mainly photographed as first images by a camera 4. The light from a second light source 3 is reflected from the outer surface of the knurling 10 of the bottom part of the bottle and, therefore, the second image only of the knurling 10 containing no foreign matter image is taken by the camera 4. When the first images are compared with the second image, the image of the knurling 10 can be discriminated from the foreign matter image and the presence of the foreign matter can be judged. When filter processing for removing a low frequency component is applied to the first images, the emboss or scratch image (noise) 11 of the body part of the bottle is removed from the first images and is not erroneously confirmed as the image of the foreign matter.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a bottle inspection method for detecting foreign matter present at the bottom of a bottle containing a substantially transparent liquid such as beer or juice using image information of the bottom of the bottle.

0002

[Prior Art] Generally, the above-mentioned bottles may have deposits left on the inside of the bottle or large scratches due to washing errors, etc., and bottles with such defects are naturally However, this not only reduces the product value but also poses a major problem in terms of food hygiene and safety. Conventionally, inspections for the presence of such defects have been carried out visually, but it is not possible to visually observe the inside of the bottle to determine the presence or absence of defects.
The results will depend on the physical condition and abilities of the inspector,
Sometimes we overlook incredibly large flaws. Since such visual inspection relies largely on human vision, it is inevitable that many defects will be overlooked.

Therefore, in recent years, various proposals have been made regarding methods and devices for automatically inspecting bottles for defects, and some are actually commercially available as empty bottle defect detectors.

These methods mainly inspect the bottle body and bottle bottom, and those that inspect the bottle body (including the side of the bottle mouth) involve shining light from one side onto the bottle being inspected, which is rotating at high speed. irradiate,
A CCD camera installed on the opposite side captures the transmitted image,
This is converted into an electrical signal, and the presence or absence of a defect is determined by an image processing device. For inspection of the bottom of a bottle, illumination is applied from below the bottom of the bottle, the transmitted image is captured by a CCD camera installed at the top of the bottle, and this signal is digitized and image processed.

[0005]

[Problem to be Solved by the Invention] However, the above-mentioned conventional bottle inspection apparatus is mainly intended for empty bottles.
The mainstream of bottle inspection after filling with liquid is to rely on visual inspection. In addition, a particularly serious defect in bottles after filling with liquid is when foreign matter gets mixed into the liquid or adheres to the bottle, and if it is present on the bottom of the bottle, it can be detected even by visual inspection. It is difficult to confirm.

When carrying out an automatic inspection of the bottom of the bottle after filling with liquid, the image taken from the TV camera above the bottle shows that the crown and
Due to stoppers, etc., it is not possible to obtain an image of the entire bottom of the bottle, and in order to inspect defects at the bottom of the bottle by imaging them with a TV camera, etc. on the side of the bottle body, image processing using several cameras is required, which is costly. The inspection will take time.

[0007] For this reason, illumination is performed from above or from the side of the bottle, and image input is performed using a TV camera installed below the bottle. Generally, bottles for beer, juice, etc. have embossments and scratches on the sides (recycled bottles are used repeatedly, so the bottles rub against each other on the conveyor during the manufacturing process, resulting in scratches), and the bottom of the bottle has scratches. Because there are depressions called knurling in the bottle, images taken with a TV camera below the bottle with lighting from above or from the side of the bottle may include shadows and bright areas caused by refraction, total internal reflection, and diffused reflection of light. As a result, images of scratches and knurling are generated, making it difficult to distinguish them from foreign objects in the bottle, and processing is required to remove these effects.

SUMMARY OF THE INVENTION Therefore, an object of the present invention is to solve the above-mentioned conventional problems, and to remove foreign matter existing in the bottom of empty bottles and liquid-filled bottles by removing scratches on the bottle body and knurling on the bottom of the bottle. The object of the present invention is to provide a bottle inspection method that allows accurate detection without being affected.

[0009]

[Means for Solving the Problems] In order to achieve such an object, the present invention provides an imaging means for receiving at least one of transmitted light and reflected light from the bottom of a bottle and converting the received light into an image signal. is disposed below the bottle bottom, and by lighting at least the first light source that irradiates light only to the lower half of the bottle body from above the bottle bottom, the imaging means A first image of the inner and outer surfaces of the bottle bottom and the foreign matter inside the bottle is acquired, and by lighting only a second light source that irradiates light toward the bottle bottom, the image of only the outer surface of the bottle bottom is captured from the imaging means. A second image is acquired, and the image processing means performs filter processing for removing low frequency components on the acquired first image to remove noise from scratches on the bottle body from the first image. The image of the foreign object in the bottle is distinguished from the image of the bottle bottom itself by comparing the second image acquired from the imaging means with the first image from which the scratch noise has been removed, and the foreign object is It is characterized by determining the presence or absence of.

0010

[Operation] In the present invention, a part of the light emitted from the first light source passes through the bottle bottom while being refracted and reflected in the bottle body and inside the bottle, and enters the imaging means. The foreign object inside the container and bottle is imaged. On the other hand, the light emitted from the second light source is reflected by the outer surface of the knurling portion at the bottom of the bottle. Therefore, when the bottle bottom is imaged using only the second light source, an image of only the knurling portion of the bottle bottom is obtained. Since the image of the knurling area is included in the first image and the second image, and the foreign object image is not included in the second image, the knurling area image and the foreign object image can be determined by comparing the first image and the second image. The presence or absence of a foreign object image can be determined. Furthermore, by applying a low frequency component removal filter process to the first image acquired using the first light source, embossments and scratch images (noise) on the bottle body are removed from the first image. The image will not be mistakenly recognized as a foreign object image.

[0011]

Embodiments Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

First, a bottle inspection device to which the present invention is applied will be explained with reference to FIGS. 2 and 3.

As shown in the figure, the bottle 1 to be inspected is conveyed while being held by a rotating wheel 6 and a bottle support plate 7, and when the bottle 1 to be inspected is positioned above the imaging means (imaging camera 4), an image of the bottom of the bottle is captured. is imaged by the imaging camera 4, and then the image processing device 5
The system is configured to perform inspection using the following methods.

At the imaging position, as shown in the figure, a light shielding plate 8, a first light source 2, and a second light source 3 are installed above the imaging camera 4, and when the bottle 1 to be inspected comes to the inspection position, the bottom of the bottle is illuminated. The light shielding plate 3 is for preventing the light from the first light source 2 from directly entering the imaging camera 4.

Here, the positions of the first light source 2 and the imaging camera 4 are set as follows. The light emitted from the first light source 2 enters the bottle 1 to be inspected and is refracted and transmitted by the bottle surface, the interface between the bottle and the internal solution, etc., but this refracted and transmitted light does not enter the lens of the imaging camera 4. installed in position. Therefore, the image captured by the imaging camera 4 basically consists of diffusely reflected light due to foreign matter in the internal solution, knurling, scratches on the bottle surface, etc.

[0016] This test attempts to detect foreign substances in the internal solution from images created by such diffusely reflected light. FIG. 4 shows an image of the bottom of the beer bottle obtained by the above method. The figure captures defects 9, knurling 10, and scratch noise 11 in the internal solution.

[0017] Here, there is a scratch on the side surface of the body of the beer bottle caused by the contact between the bottles, and when this part is irradiated with light from the first light source 2, the light diffusely reflected from the bottle surface is reflected on the bottle. The noise is transmitted through the interior by total reflection and appears as scratch noise 11 as shown in FIG.

[0018] Since these scratches occur almost all around the circumferential direction of the bottle, the image is long in the circumferential direction as shown in the figure. When recognizing the defect 9 from such an image, it is necessary to distinguish it from a foreign object image as scratch noise.

[0019] In such a case, the judgment may be based on the magnitude of the brightness value of the output image, but in the case of a bottle that has been used repeatedly, the scratch noise 9 is the same or higher in brightness level than the defect signal 8. In many cases, simply comparing the luminance values is not practical. Therefore, in this embodiment, the following processing is carried out by the image processing device by utilizing the fact that the continuous change in brightness in the circumferential direction of the image of such scratch noise is gentler than that of the defect image.
to remove scratch noise.

First, in order to explain the method of inputting the luminance values of pixels to the image processing device 5 along the circumferential direction, an example of an inspection image is shown in FIG.

[0021] The squares in the figure indicate pixels, and the pixel positions are indicated in an orthogonal coordinate system. In the case of this process, address values of squares along the circumferential direction are stored in advance in the memory of the image processing device 5 in order to perform low frequency component removal filter processing, which will be described later, in the circumferential direction of the bottle. It is assumed that the luminance values of each pixel in the circumferential direction according to this address are a1 to ak. Assuming that the brightness value of the current pixel of interest is ai, the scratch noise is removed by correcting the brightness value of the pixel of interest by low frequency component removal filter processing using the following equation.

[0022]

[Math 1]

[0023] This filter processing calculates the average of the luminance levels of 2N+1 constant lengths centered on the pixel of interest, and then multiplies the average value by x times (reference value) from the luminance level of the pixel of interest. It means to decrease. Here x is 0.3
The value is around ~5, and when the value is set to a large value, the effect of erasing areas where high-brightness pixels are continuously present becomes stronger, and scratch noise can be easily removed. In the case of a defect, it will be erased in the same way as scratch noise, so a value of about 0.5 to 2 is appropriate.

This filter processing will be explained in more detail with reference to FIG. FIG. 7 shows the distribution of brightness values in the circumferential direction of the bottle in the image signal taken by the imaging camera 4, and the brightness of an image obtained by applying a low frequency component removal filter process to the brightness distribution as shown in Equation 1. This shows the distribution. Here, the x value in equation 1 is set to "1". Although FIG. 7 shows an image signal in a portion without knurling for easier explanation, the effect is the same even in a portion where knurling exists.

The upper waveform in FIG. 7 shows the brightness distribution before filter processing, and as shown in this distribution, the signal of the defective part changes rapidly, whereas the brightness of the scratch noise changes gradually. I can see that it is changing. Therefore, in the scratched portion, there is no large difference between the brightness level of the pixel of interest and the average value, and the brightness level ai′ after correction becomes a value close to zero.

On the other hand, the brightness of the defective pixel changes rapidly at the outer edge of the defect, and the constant length range for filtering (2N+1 in this example) is larger than the size of the defect.
The luminance level of the pixel of interest is greater than the average value.

Therefore, as shown in the lower waveform of FIG. 7, the brightness value of the defective part becomes larger than that of the scratched part due to the filtering process, and it can be easily distinguished from scratch noise.

[0028] In this example, we have shown an example in which the range of a certain length for filter processing is larger than the defect size, but even if the defect size is larger, the brightness at the outer edge changes rapidly. Therefore, the brightness value of the outer edge portion becomes a value sufficiently larger than the average value, making it possible to distinguish it from scratch noise.

Next, a second stage of image processing is performed to erase the knurling portion. The image obtained by the above-mentioned first light source and subjected to filter processing is used as the first image, and the bottle to be inspected 1 that has been guided to the inspection position in the bottle inspection apparatus shown in FIG. The image obtained by irradiating the bottom of the bottle is defined as the second image.

[0030] Normally, beer bottles have very low glass transmittance, so the image captured using this method only represents the shape of the bottom surface of the bottle, and images of the inside of the bottle, such as foreign objects in the liquid, cannot be seen. I can't take pictures. For this reason, in the case of beer bottles, images of knurling and scratches on the bottom surface of the bottle are formed. An example of the second image is shown in FIG. This second image is binarized using a certain threshold. In this method, brightness values are expressed in 256 gradations from 0 to 255, and the portion corresponding to knurling is set to 255 by binarization.

The dilation process here is a process for enlarging the knurling area recognized in the second image, and is a method often used as an image processing method. Although this expansion process is performed for the following purposes, it is not necessarily necessary to perform it.

[0032] The knurling areas of the first image and the second image do not necessarily match because the projected illumination is different. For this reason, if the knurling image of the second image is smaller than the knurling image of the first image, it is impossible to completely eliminate the knurling using the means described below. For this reason, the second image is subjected to dilation processing and subjected to the following processing. Next, FIG. 9 shows an example of an image after binarization and expansion. By subtracting this binarized and expanded image from the filtered first image (FIG. 5), the knurling portion is eliminated. An example of this processing result is shown in the first
Shown in Figure 0.

The images of the knurling portion of the first image and the second image are slightly different, and expansion processing is performed after binarization to prevent unerased areas during subtraction processing.

Finally, after the image after the subtraction is binarized using a predetermined threshold, the black-white area ratio (if white is considered a defect, (number of white pixels)/(number of black pixels)) is calculated, The total is compared with a preset value, and if the total is equal to or greater than the threshold, the bottle is determined to be defective. Then, if necessary, a reject signal is outputted from the image processing device 5 to remove this bottle from the transport system.

[0035] Note that the description of the position detecting means provided in connection with the conveying means to obtain the timing of image capturing by the image capturing camera 4, the defective bottle ejecting means, and their operations will be omitted.

FIG. 11 shows an example of the circuit configuration of the image processing device 5 for executing such processing.

In FIG. 11, the central processing unit (CP
U) 100, read-only memory 110, random access memory (RAM) 120, keyboard input device 13
0, display device 140, analog-digital (A/D) converter 150, digital-analog (D/A) converter 16
0 is connected to the common bus. The CPU 100 reads the ROM in response to an execution command input from the keyboard input device 130.
The control procedures shown in FIGS. 1 and 12 stored in 110 are executed. Note that FIG. 12 shows the process of S30 in FIG. 1 in more detail.

Two types of image signals (analog format) output from the imaging camera 4 as imaging results are sent to the A/D converter 1.
50, the signal is converted into a signal indicating the brightness level of each pixel in digital form, and is written into the RAM 120 by the CPU 100 (S10, S20 in FIG. 1).

[0039] The CPU 100 selects R from the RAM 120.
Based on the address table recorded in the OM110, the luminance data of pixels in the circumferential direction is read out, the above-mentioned filter processing is performed, and the luminance data after filter correction is stored in the RAM.
120 (steps S31 to 120 in FIG. 12)
loop processing in S34). After finishing filtering,
The CPU 100 stores the second image in the RAM 120 and the ROM 1 in advance.
Binarization is performed by comparison with a threshold value A set to 10 (step S40 in FIG. 1), and the result is written into the RAM 120.

Thereafter, the CPU 100 performs subtraction to compare the first image and the second image, and the result is set in the ROM 110 in advance. Submit a defect image having a value larger than threshold B (step S60 in FIG. 1)
. The CPU 100 counts the number of pixels of this defective image,
Bottles for which this number exceeds a certain level are determined to be defective bottles (steps S70 and S80 in FIG. 1).

The present invention is not limited to this embodiment, and the present invention can be implemented in various ways as follows.

1) In this embodiment, the range for averaging is 8 degrees in the circumferential direction, but it is effective if it is 2 degrees or more. As the range of this averaging process becomes larger, defect signals can be detected with larger values, but the effect of erasing scratch noise becomes smaller. On the other hand, if the range of the averaging process is made smaller, the scratch noise becomes smaller, but the defect signal also becomes smaller. The range of averaging processing should be set to an optimal value depending on the type of bottle, and is generally 2 to 90 degrees in the circumferential direction.
degree is desirable.

2) In this embodiment, when extracting pixels in the circumferential direction, the brightness value of each pixel was read out according to the address stored in the memory. may be obtained by calculation. In addition, although the brightness value of a certain address was used as is, in order to obtain a more accurate brightness value in the circumferential direction, it is also possible to calculate the brightness value along the circumferential direction by interpolation from the brightness values of neighboring pixels. be.

3) In this embodiment, only the scratch noise on the bottom of the bottle is removed, but when the bottom of the bottle has an embossment, the noise image caused by the embossment can also be removed. In this case, filter processing is performed on image signals for one screen.

4) In this embodiment, the second image is used for the subtraction process after the binarization/expansion process, but it is also effective to perform the subtraction process as it is the captured image.

5) In this example, one imaging camera was used to take the first image and the second image in order, but it is also possible to configure a half mirror in the optical path and take the images with two cameras. do not have.

The details of this are omitted here because the applicant has already filed an application, but by using this method and further illuminating with a strobe, it is possible to capture the first and second images almost simultaneously. Furthermore, since there is no need to keep the bottle still during transport, high-speed processing becomes possible.

[0048]

As explained above, according to the present invention, scratch noise on the bottle body can be automatically removed from the first image used for bottle inspection, and foreign matter inside the bottle and the bottom of the bottle can be removed automatically. Since the first image and the second image of only the bottle bottom are compared, it is possible to distinguish between the bottle bottom image and the foreign object image. As a result, scratch noise in the captured image and knurling images at the bottom of the bottle are not mistakenly recognized as foreign objects, and the accuracy of foreign object detection is increased.

[Brief explanation of the drawing]

FIG. 1 is a flowchart showing a bottle inspection procedure in an embodiment of the present invention.

FIG. 2 is a schematic diagram showing the structure of a bottle inspection device for implementing the present invention.

FIG. 3 is a schematic diagram showing the structure of a bottle inspection device for implementing the present invention.

FIG. 4 is an explanatory diagram showing an image before filter processing in the embodiment of the present invention.

FIG. 5 is an explanatory diagram showing an image after filter processing in the embodiment of the present invention.

FIG. 6 is an explanatory diagram showing address values for referring to the luminance values of pixels in the circumferential direction from the bottle bottom image in the embodiment of the present invention.

FIG. 7 is a waveform diagram showing the luminance distribution before and after the low frequency component removal filter in the embodiment of the present invention.

FIG. 8 is an explanatory diagram showing an example of a second image in the embodiment of the present invention.

FIG. 9 is an explanatory diagram showing an example of a second image after binarization and expansion in the embodiment of the present invention.

FIG. 10 is an explanatory diagram showing the result of subtracting the binarized second image from the first image in the embodiment of the present invention.

11 is a block diagram showing a circuit configuration of the image processing device 5 of FIG. 3. FIG.

FIG. 12 is a flowchart showing the procedure of low frequency component removal filter processing according to the embodiment of the present invention.

[Explanation of symbols]

1 Bottle to be inspected 2 First light source 3 Second light source 4 Imaging camera 5 Image processing device 6 Rotating wheel 7 Bottle support plate 8. Light shielding plate 9,9’,9” Defect 10,10’,10” Knurling 11 Torso abrasion noise 12. Scratch on the bottom of the bottle 100 CPU 110 ROM 120 RAM 130 Keyboard input device 140 Display screen 150 A/D converter 160 D/A converter

Claims (1)

[Claims] 1. Imaging means for receiving at least one of transmitted light and reflected light from the bottle bottom and converting the received light into an image signal is disposed below the bottle bottom, and the imaging means is arranged below the bottle bottom to receive and convert at least one of transmitted light and reflected light from the bottle bottom to an image signal. from the imaging means by turning on at least the first light source that irradiates light only to the lower half of the bottle body;
A first image of the inner and outer surfaces of the bottle bottom and the foreign matter inside the bottle is acquired, and by lighting only a second light source that irradiates light toward the bottle bottom, the imaging means captures only the outer surface of the bottle bottom. Obtaining the second image and using an image processing means,
A filter process for removing low frequency components is applied to the acquired first image to remove scratch noise on the bottle body from the first image, and a second image acquired from the imaging means is obtained. , the first one with the scratch noise removed
A bottle inspection method characterized in that the image of the foreign object in the bottle is distinguished from the image of the bottle bottom itself by comparing the images, and the presence or absence of the foreign object is determined by comparing the images.