JPH0122896B2 - - Google Patents

Description

[Detailed description of the invention] [Industrial application field]

The present invention relates to a method for inspecting a container having a smooth surface such as a beer bottle, and in particular, for objectively evaluating the degree of scratches (scratches) generated on the surface thereof, and sorting out good products from defective products, and/or It relates to a method of grading and sorting according to the degree of scathing.

[Conventional technology]

(Background) Today, extremely large quantities of glass bottles are in circulation as containers for beer bottles, soft drink bottles, Western liquor bottles, stored foods, chemicals, and other items. Conventionally, it has been customary to collect and reuse some of these items, such as beer bottles and ramune bottles, after the contents have been consumed. In addition, it is thought that other bottles that are not currently being collected will need to be collected in the near future from the standpoint of resource conservation. However, collected bottles always have more or less minute scratches (scuffs) during the process of collection, transportation, cleaning, and conveyor transfer within the factory, and if they are reused, they will be seriously damaged. Reduce product value. For this reason, manufacturers who reuse the bottles have a habit of inspecting the condition of the scats before reusing them, and excluding those that are in poor condition. (Problems with the prior art) However, this test has so far been performed exclusively with the naked eye, which requires considerable skill. Fluctuations in results due to differences and fatigue cannot be ignored, and in particular, for colorless bottles such as carbonated water, there are problems such as variations in test results can affect product sales. Furthermore, since visual inspection relies on delicate subjective judgments, it is extremely difficult to grade the degree of skatosis, and for this reason, it is virtually impossible to sort bottles according to the purpose. Therefore, although some attempts have been made to mechanize this work, it is not easy to obtain results similar to those obtained by visual inspection, and as far as the inventors know, none has yet been put to practical use.

Problems to be Solved by the Invention The present invention solves the problem of errors caused by individual differences and fatigue in the conventional visual scuff inspection by providing a mechanical sorting means that closely matches the results of visual inspection. The purpose is to solve the problem.

[Means to solve the problem]

(Background of the invention) By the way, the essence of the visual action of the human eye can, in principle, be attributed to the form of pulses emitted by rod-shaped cells, which are responsible for photopic vision and color vision, and cone-shaped cells, which are responsible for scotopic vision. However, as a result of complex integration actions taking place in the visual center, the actual action is extremely complex, and it is currently impossible to clarify its true nature. However, the present inventor discovered that among the pulse voltages transmitted from the optic nerve based on the difference in brightness between the Skatuff part and other parts, the integrated value of the voltage exceeding a certain threshold is compared with a standard value that exists in the central memory in advance. Assuming that the number of pulses exceeds a certain set threshold (noise level), we observe the beer bottles using an image sensor camera and rank them according to the magnitude of the integrated value of the number of pulses exceeding a certain set threshold (noise level). When compared with the above visual ranking, it was discovered that there was a significant correspondence between the two as shown in the table below, and it was verified that the inventor's hypothesis was substantially true. Incidentally, this data is
Using an image sensor camera with a photodiode array with 512 elements, the scanning frequency is 50
Hz, camera aperture F4, threshold line (noise level)
4V, halogen lamp light source, camera - center of bottle -
These are the results of experiments conducted at eight locations around the bottle with an included angle of 60° between the light sources.

[Table] *Mechanical observation results are also arranged according to the visual observation order.As can be seen from the table above, the order of the number of bits according to the visual observation method and the order of the number of bits according to the machine observation method are somewhat different, but As can be seen from the above results, the rankings determined by the visual method and the rankings determined by the mechanical method are somewhat different, but they match well after around 100 bits, which is considered to be the sorting line. The inspection method based on the concept of
It can be seen that this is close to human vision. Incidentally, when calculating Spearman's rank correlation coefficient from the above results, we obtain the value γ _S = 1-6Σdi ² /n ³ - n = 0.93 (di: difference in rank, n = number of samples), which is When tested, it is highly significant (significance level 1%). In other words, there is a strong correlation between the visual method and the electro-optical measurement method, which uses an image sensor camera and integrates the number of pulses exceeding a threshold, and the latter can sufficiently replace the former. It is shown that. (Summary) The present invention is based on the above knowledge, that is, an experiment in which the magnitude of the integral value of the number of video impulses captured by an image sensor camera is correlated with the degree of scattering of light by a visual inspection method. Based on the results, a light beam is irradiated onto the body of the object to be inspected, such as a glass bottle, placed upright on a rotating table, and among the reflected light that is reflected during one revolution of the object,
The scattered light contained in the reflected light in the direction making an acute angle with the incident axis of the irradiation light beam is converted into a pulsed video signal using an image sensor type imaging tube, and the converted video signal pulse exceeds a noise level. The gist of this is a method for sorting glass bottles, etc., characterized by comparing the integral value of the signal pulse number with a standard integral pulse number determined from visual observation, and excluding objects to be inspected that exceed the standard integral value. It is. (Object to be inspected) The term "glass bottle, etc." is used here as an example since glass bottles are generally a typical object to be inspected, and there is no need to refer to it as a "bottle."
Naturally, it can also be applied to vessels such as anchor containers and ampoules. In addition, the material may be plastic. Furthermore, it is irrelevant whether the material is transparent or colored, transparent or opaque. However, if the surface has significant irregularities or is extremely roughened, the reflection level will not be leveled, and therefore it is not suitable as a subject of the invention.
The best object for the present invention is one with a smooth surface. (Apparatus) A halogen lamp is preferably used as the light source, but other light sources such as natural light, an ordinary incandescent lamp, a fluorescent lamp, a mercury lamp or a Zenon lamp can also be used. The image pickup tube used for light reception (imaging) is CCD.
A self-scanning image sensor camera using a charge-coupled device (charge-coupled device) element or an integrated photodiode array as a light-receiving element is most suitable in terms of cost, durability, error, etc. The camera should be positioned so that it can capture the reflected light in a direction where the angle is acute with respect to the line connecting the light source and the axis of the glass bottle to be inspected, or in other words, in a direction which is acute with the incident axis of the irradiated light beam. Placed. In the case of bottles and the like, scats are rarely present evenly over the entire circumference, but rather are often locally unevenly distributed. Therefore, the detection of skats is performed by rotating the object once around its axis, or conversely, by rotating a set of light source and camera around the body of the object. For inspection purposes, the former method of applying rotation to the object itself is appropriate. In this case, the object is placed upright on a conveyor equipped with a rotary table and is moved intermittently and rotated during periods of stoppage, during which measurements are taken. Rejection of rejected objects whose integrated value of the number of pulses exceeding the threshold value exceeds the standard can be easily carried out using, for example, a solenoid, an air cylinder with a solenoid valve, or the like. It is preferable to place a black screen on the back of the object facing the light source to avoid reflection of light and intrusion of outside light. (Threshold value and standard value) When actually measuring with a video camera, various background noises are generated due to the influence of electronic circuits, etc., so when measuring, it is necessary to cut the noise at a certain level (threshold voltage) and make sure that it is not related to the actual scuff. It is necessary to remove low-level pulses that seem to be. Although this noise varies depending on its value, the object to be inspected, the measurement conditions, etc., the value of 4V is an average value derived from the results of many experiments. The standard value is determined by comparing the number of integrated pulses (number of bits) obtained by mechanical observation, excluding weak pulses below the noise level, with the results of visual observation, as in the above experiment. The upper limit is the average number of integral bits that is considered inconspicuous. For a normal smooth glass bottle, the number of bits is about 100 under the conditions of the experiment described above, but if you want to select a specially selected bottle, such as a ceremonial bottle, the number of bits must be lowered. Of course.

[Effect]

The present invention can contribute to labor saving by making it possible to automate bottle inspection, but its greatest significance is that it can completely eliminate deviations due to individual differences and fatigue. Furthermore, since it is easy to grade, which is impossible with visual inspection, it is possible to accurately sort specially selected bottles such as liquor bottles for weddings, celebrations, other celebrations, gifts, luxury hotels, private use, etc. It brings about effects such as being able to be implemented.

【Example】

The drawing shows a schematic representation of an example of a device for carrying out the method of the invention. Beer bottles (1, 1...) to be inspected are placed on a conveyor with a rotating table (2, 2...) and successively transported along a horizontal plane in the direction of the arrow. The light source 3 and camera 4 are arranged so that the included angle (θ) with the long axis of the bottle at the inspection position is about 60°. At the inspection position, the bottle 1 is rotated, for example, in the direction of the arrow, and rotates once within a short period of one second (although it may be slower). During this rotation, the light captured by the image sensor camera 4 is
It is converted into pulses in the camera, and only the number of pulses exceeding the set value is taken out. This number of pulses is integrated by the calculator 5. At the same time, for bottles whose integrated value exceeds the standard value, a removal device (Fig. A signal is sent to start the operation (not shown). The number and strength of the pulses are determined by the waveform 7 using the synchroscope 6.
It can be visually observed as

【Effect of the invention】

As explained above, according to the present invention, it is possible to automatically and accurately perform the inspection of the scats, which has so far relied on visual inspection, and therefore, highly reliable quantitative results can be obtained without the intervention of subjectivity or fatigue. In addition, it is possible to grade the degree of scuffing, and the inspection speed is much higher than that of visual inspection, so it can contribute to streamlining bottle selection work.

[Brief explanation of drawings]

The drawings are schematic perspective views for explaining an apparatus for carrying out the method of the present invention and its operation. The attached reference photo shows a situation in which test bottles, numbered by visual inspection in descending order of scuffs, are arranged in descending order of the number of bits observed by an image sensor camera. The meanings of the symbols in the diagram are as follows: - 1: Glass bottle, 2: Conveyor, 3: Light source, 4:
Image sensor camera; 5: Calculator;
6: Synchroscope; 7: Waveform, 8: Threshold line, 9: Light absorption screen.

Claims (1)

[Claims] 1. A light beam is irradiated onto the body of an object to be inspected, such as a glass bottle, placed upright on a rotating table, and among the reflected light that is reflected during one revolution of the object, the incident axis of the irradiated light beam is Convert the scattered light contained in the reflected light in the direction making an acute angle to the pulse type video signal using an image sensor type imaging tube, and integrate the number of signal pulses exceeding the noise level among the converted video signal pulses. A method for sorting glass bottles, etc., characterized in that the value is compared with a standard integral pulse number determined from visual observation, and objects to be inspected that exceed the standard integral value are excluded.