JP5390652B2 - Non-defective product judgment standard setting method and non-defective product judgment standard setting device in inspection apparatus - Google Patents

Description

  The present invention executes a non-defective product criterion setting method for setting a non-defective product criterion for determining whether a test object is a non-defective product or a non-defective product in an inspection device that determines the quality of the test object. The present invention relates to a non-defective product criterion setting device.

  2. Description of the Related Art Conventionally, an inspection apparatus that determines the quality of an inspection object based on an inspection measurement value obtained from the inspection object has been widely used. For example, an image inspection apparatus that determines the quality of an inspection object based on an inspection measurement value obtained from an image signal captured by an imaging means that captures an image of the inspection object has been widely used for quality inspection of food containers and the like. Has been.

  In general, an inspection object is imaged by an imaging means such as a CCD camera, the image signal is digitized, and the image is binarized with an appropriate threshold value. This determination is performed.

Japanese Patent Laid-Open No. 10-148515

  When determining whether a non-defective product is defective or defective in an inspection device such as the image inspection device described above, for example, when determining whether a non-defective product is defective or defective from the number of pixels in the image inspection device, a non-defective product determination criterion is set in advance. In general, an inspection object that does not satisfy the non-defective product determination standard is regarded as a defective product.

As a method of developing this and setting good product judgment criteria, it is assumed in advance that multiple good product measurement values obtained using the inspection device are expressed in a normal distribution for multiple good products. _MG and non-defective product standard deviation σ _G are obtained to obtain non-defective product weighing value distribution, and the non-defective product average value _MG ± k × non-defective product standard deviation σ _G (k is a positive number) is used as the non-defective product judgment standard. A method of certifying a product that does not satisfy the requirements as a defective product is conceivable.

For example, when the inspection device is an image inspection device, it is assumed that a plurality of non-defective product weight values obtained from the image signal captured by the imaging means are represented by a normal distribution, and the non-defective product average value _MG and the non-defective product standard deviation seeking a sigma _G, to obtain a good metric value distribution as shown in FIG. 6 shown, good average value M _G ± k × good standard deviation sigma _{G (k} is a positive number) is determined as good determination reference, the non-defective criteria This is a technique for certifying unsatisfied products as defective products.

When the non-defective product determination criteria are set as described above, for a plurality of defective products, an inspection device, for example, an image inspection device is used, and the plurality of defective product measurement values obtained from the image signal captured by the imaging unit are normally distributed. As shown in FIG. 7, the non-defective product measurement value distribution obtained by calculating the average value M _NG of defective products and the standard deviation σ _NG of defective products is shown in FIG. Assuming that they are sandwiched, the portions denoted by reference numerals 103 and 104 in FIG. 7 are in a range in which there is a possibility that a defective product is generated even though the non-defective product determination criteria are satisfied.

  In this case, as described above, after accepting that there may be a non-defective product that is recognized as a “defective product” and rejected, a product that does not satisfy the non-defective product determination criteria. Despite being recognized as “defective” and rejected uniformly, defective products will be mixed into products that have been certified as good, that is, “defective products will be judged as good. (= Determining defective product is defective).

  Such “defect product determination” is handled as an inspection error. Inspection errors occur only with the possibility of about once every tens of thousands of times. For example, when the object to be inspected is food that is eaten or eaten by humans, the ratio is one in every tens of thousands. But if an inspection error occurs, it becomes a serious problem.

Therefore, efforts to improve inspection accuracy are made every day, but as described above, it is assumed that a plurality of non-defective product weight values obtained in advance using an inspection device are represented by a normal distribution, and the non-defective product average value is obtained. _MG and non-defective product standard deviation σ _G are obtained, the non-defective product weight value distribution is obtained, and the non-defective product average value _MG ± k × non-defective product standard deviation σ _G (k is a positive number) is used as a non-defective product judgment standard. It is impossible to avoid the occurrence of defective products such as the areas indicated by reference numerals 103 and 104 described with reference to FIG.

  In addition, the conventional inspection apparatus has a function capable of performing self-diagnosis in the course of performing an inspection as to whether or not the inspection apparatus (for example, an image inspection apparatus) continues to maintain the accuracy of the non-defective product determination criteria once set. It was not done.

  For example, in an image inspection apparatus, the amount of light caused by deterioration of the light source of a CCD camera for capturing images over time, lens dirt, CCD camera attachment, and the conveyor that conveys the inspection object Judgment accuracy may fluctuate due to knocking, misalignment of guides, noise of communication cables connecting various devices, etc., but conventional image inspection devices are subject to these disturbances in the course of ongoing inspection. Therefore, no function has been provided that can determine whether or not the determination accuracy varies, that is, whether or not the determination accuracy is maintained.

  The present invention is obtained from an inspection apparatus that determines the quality of an inspection object based on an inspection measurement value obtained by using the inspection apparatus for the inspection object, for example, an image signal captured by an imaging unit that captures an image of the inspection object. A non-defective product criterion setting method for setting a non-defective product criterion for determining whether a test object is a non-defective product or a non-defective product in an image inspection device that determines the quality of a test object based on the measured value of inspection. , “Determining a defective product as a non-defective product”, that is, a non-defective product determination that can achieve the purpose of quality control of eliminating “defective product determination” (= statistically zero) It is an object of the present invention to provide a reference setting method and a pass / fail judgment reference setting apparatus in an inspection apparatus.

  In the present application aiming at achieving the above object, first, a non-defective product determination standard setting method for an inspection apparatus is described. This is a non-defective product criterion setting method for setting a non-defective product criterion for determining whether a test object is a non-defective product or a defective product in an inspection device that determines the quality of the test object. Obtain the distribution of non-defective product weight values obtained using the inspection device and the distribution of non-defective product weight values obtained using the inspection device for a plurality of defective products, and set the non-defective product judgment criteria by comparing the two. It is.

  For example, as shown in FIG. 1, in the non-defective product weight value distribution (normal distribution curve 40), the upper limit non-defective product determination criterion is assumed to be smaller than the lower limit of the defective product weight value distribution (normal distribution curve 42) on the upper limit side. On the other hand, the non-defective product criterion is set to be larger than the upper limit of the distribution of defective product measurement values (normal distribution curve 41) on the lower limit side.

  In this way, compared with the case where the non-defective product judgment criterion is set only based on the non-defective product weight distribution as described with reference to FIGS. There is a possibility that the number of items that are judged to be non-defective products may increase, but "determining defective products as good products", that is, "determining defective products as good products" may be made statistically zero. It becomes possible.

  A non-defective product criterion setting apparatus in an inspection device in which such a non-defective product criterion setting method is executed determines whether the inspection object in the inspection device that determines the quality of the inspection object is a good product or a defective product. A non-defective product determination criterion setting device for setting a non-defective product determination criterion, and a non-defective product weight value distribution generating means for obtaining a distribution of a plurality of good product weight values obtained by using the inspection device for a plurality of good products, and a plurality of defective products A defective measurement value distribution generating means for obtaining a distribution of a plurality of defective measurement values obtained using the inspection apparatus, a non-defective measurement value distribution obtained by the non-defective measurement value distribution generating means, and a defective measurement value And a first distribution state image display means for comparing and displaying the distribution of defective product measured values obtained by the distribution generation means on the same screen.

  By the first distribution state image display means in the non-defective product judgment reference setting device of this inspection device, the non-defective product weight value distribution and the defective product weight value distribution on the same screen are, for example, as shown in FIGS. Therefore, for example, in the distribution of the non-defective product measurement values (normal distribution curve 40), the upper non-defective product determination standard is set to the distribution of the defective product measurement values on the upper limit side (normal distribution curve 42). The non-defective product criterion can be set by setting the non-defective product criterion below the lower limit to be larger than the upper limit of the distribution of defective product measurement values (normal distribution curve 41) on the lower limit side.

  In this way, compared with the case where the non-defective product determination standard is set based only on the distribution of the non-defective product measurement values as described with reference to FIGS. Even if there is a possibility that the number of items to be judged will increase, it will be possible to statistically make "determined a defective product as a non-defective product", that is, "determine a non-defective product as a non-defective product". .

  The above-described non-defective product judgment criterion setting method for an inspection device and the non-defective product judgment standard setting device for an inspection device are inspected based on an inspection measurement value obtained from an image signal captured by an imaging device that captures an image of an inspection object. It can be employed in the case of an image inspection apparatus that determines the quality of an object.

Further, the present application describes another non-defective product criterion setting method for the inspection apparatus. This is a non-defective product criterion setting method for setting a non-defective product criterion for determining whether a test object is a non-defective product or a defective product in an inspection device that determines the quality of the test object. Assuming that a plurality of non-defective product weight values obtained by using the inspection device are represented by a normal distribution, a non-defective product average value _MG and a non-defective product standard deviation σ _G are obtained, and a plurality of defective products are inspected. Assuming that a plurality of defective product measurement values obtained using the apparatus are represented by a normal distribution, an average value M _NG of defective products and a standard deviation σ _{NG of} defective products are obtained, and the distribution of non-defective product measurement values and The non-defective product determination standard is set by comparing with the distribution of non-defective product measurement values.

Here, the step of setting the non-defective product determination standard by comparing the distribution of the non-defective product measurement value and the distribution of the defective product measurement value is, for example, M _NG ± kσ _NG (k is a positive number in the distribution of the good product measurement value). Thus, preferably, an inspection measurement value corresponding to k = 3, 4, 5 or 6, more preferably k = 6) can be set as a non-defective product determination standard.

Also, the process of setting the non-defective product judgment value by comparing the non-defective product measure value distribution with the defective product measure value distribution is the same screen for the non-defective product measure value distribution and the non-defective product measure value distribution. 1 and 3 can be displayed, and more preferably, M _G ± kσ _G (k = 1, 2, 3, 4, 5 or 6) is displayed on the same screen. The range and the range of M _NG ± kσ _NG (k = 1, 2, 3, 4, 5 or 6) are displayed, and the range is color-coded according to k. be able to.

Then, the non-defective product determination criterion setting device in which the above-described non-defective product determination criterion setting method is implemented is a non-defective product determination criterion for determining whether the inspection object in the inspection device that determines the quality of the inspection object is a good product or a defective product. And a non-defective product average value _MG and a non-defective product assuming that a plurality of non-defective product weight values obtained by using the inspection device are represented by a normal distribution. Assuming that, for a plurality of defective products, a plurality of defective product measured values obtained by using the inspection device are expressed in a normal distribution with respect to a first non-defective product average value / standard deviation calculating means for calculating the standard deviation σ _G The first defective average value / standard deviation calculating means for calculating the defective average value M _NG and the defective standard deviation σ _NG , the non-defective product weight value distribution, and the defective product weight value distribution The first distribution to display on the same screen for comparison It is a good criterion setting device in the test apparatus characterized by comprising an image display unit.

By the first distribution state image display means in the non-defective product judgment reference setting device of this inspection device, as shown in FIG. 1 and FIG. Compared with the distribution, for example, in the distribution of non-defective product weight values, M _NG ± kσ _NG (k is a positive number, preferably k = 3, 4, 5 or 6, more preferably k = 6 ) Can be set as a non-defective product criterion.

In addition, in this way, the distribution of non-defective product weight values and the distribution of defective product weight values displayed on the same screen are respectively M _G ± kσ _G (k = 1, 2, 3, 4, 5 or 6) and M _NG ± kσ _NG (k = 1, 2, 3, 4, 5 or 6) range are displayed, and the range is color-coded according to k. Can be done.

According to the experiments by the inventors, if an inspection measurement value corresponding to M _NG ± kσ _NG (k = 6) is set as a non-defective product determination standard in the distribution of non-defective product measurement values, “determining that a defective product is a non-defective product” It was possible to eradicate the “determination”, that is, to “determine that the defective product is non-defective” (= statistically zero).

The above-described non-defective product determination criterion setting method can perform the inspection apparatus as the above-described image inspection. In other words, the non-defective product determination criterion setting method of the image inspection apparatus described in the present application determines the quality of the inspection object based on the inspection measurement value obtained from the image signal captured by the imaging unit that captures the image of the inspection object. A non-defective product criterion setting method for setting a non-defective product criterion for determining whether an inspection object is a non-defective product or a non-defective product in an image inspection device, wherein the imaging is performed using the image inspection device for a plurality of non-defective products. Assuming that a plurality of non-defective product weight values obtained from the image signal captured by the means are represented by a normal distribution, a non-defective product average value _MG and a non-defective product standard deviation σ _G are obtained. using the inspection device, assuming that a plurality of defective weighing value obtained from the image signal captured by the imaging means is represented by a normal distribution, a defective average M _NG, defective standard deviation sigma _N Preparative determined, it is characterized in that to set the distribution of good metric, the non-defective criteria by comparing the distribution of the defective weighing.

Here, as described above, the step of setting the non-defective product determination standard by comparing the distribution of the non-defective product measurement value and the distribution of the defective product measurement value is, for example, M _NG ± kσ _NG ( k is a positive number, and preferably, an inspection measurement value corresponding to k = 3, 4, 5 or 6, more preferably k = 6) can be set as a non-defective product determination criterion.

In addition, the process of setting the non-defective product determination standard by comparing the distribution of the non-defective product measurement value with the distribution of the non-defective product measurement value includes the distribution of the non-defective product measurement value and the measurement value of the defective product as shown in FIGS. Can be displayed and displayed on the same screen, and more preferably M _G ± kσ _G (k = 1, 2, 3, 4, 5 or 6) on the same screen. And a range of M _NG ± kσ _NG (k = 1, 2, 3, 4, 5 or 6) are displayed, and the ranges are color-coded according to k, for example, to compare the two. Can be.

A non-defective product determination standard setting apparatus in which such a non-defective product determination standard setting method is implemented determines the quality of an inspection target based on an inspection measurement value obtained from an image signal captured by an imaging unit that captures an image of the inspection target. A non-defective product determination criterion for determining whether an inspection object in the image inspection apparatus is a non-defective product or a defective product is obtained. An image captured by the imaging unit using the image inspection device for a plurality of non-defective products. assuming that a plurality of non-defective weighed value obtained from the signal is represented by a normal distribution, a good average value M _G, and a second good average, standard deviation calculation means for calculating a non-defective standard deviation sigma _G, for a plurality of defective products, using the image inspection device, assuming a plurality of defective weighing value obtained from the image signal captured by the imaging means is represented by a normal distribution, a defective average M _NG A second defective mean, standard deviation calculation means for calculating a defective standard deviation sigma _NG, the display by comparing the distribution of good metric, and a distribution of the defective weighing on the same screen And a second distribution state image display means.

The second distribution state image display means in the non-defective product judgment criterion setting device of this image inspection apparatus compares the distribution of the non-defective product weight value displayed on the same screen with the distribution of the defective product weight value, Similarly, for example, in the distribution of non-defective product weight values, it corresponds to M _NG ± kσ _NG (k is a positive number, preferably k = 3, 4, 5 or 6, more preferably k = 6) The inspection measurement value to be set can be set as a non-defective product judgment standard.

As described above, the distribution of non-defective product weight values and the distribution of defective product weight values (for example, shown in FIGS. 1 and 3) displayed on the same screen are respectively M _G ± kσ _G (k = 1, 2, 3, 4, 5 or 6) and M _NG ± kσ _NG (k = 1, 2, 3, 4, 5 or 6), and the range according to k Can be compared with each other by, for example, displaying them in different colors.

As described above, according to experiments by the inventors, if an inspection measurement value corresponding to M _NG ± kσ _NG (k = 6) is set as a non-defective product determination standard in the distribution of non-defective product measurement values, It was possible to eradicate “determining that the product is non-defective”, that is, eliminating “determining non-defective product” (= statistically zero).

  According to the non-defective product determination standard setting method of the inspection apparatus (for example, image inspection apparatus) described in the present application and the non-defective product determination standard setting apparatus of the inspection apparatus (for example, image inspection apparatus) described above, for example, a new inspection apparatus Before actually starting the inspection of the inspection object using the inspection device, the non-defective product determination criterion setting method of the inspection device described in the present application is implemented using the non-defective product determination criterion setting device of the inspection device described in the present application. By doing so, it is possible to confirm the accuracy of the non-defective product determination criteria set by the new inspection device, that is, the accuracy of the non-defective product determination by the new inspection device.

Non-defective product criteria set by the apparatus and method described in the present application (for example, M _NG ± kσ _NG (k is a positive number in the distribution of non-defective product weight values, and preferably k = 3, 4, 5 Or, more preferably, the inspection weighing value corresponding to k = 6) is compared with the non-defective product determination criteria set in the new inspection device from the beginning, thereby determining the accuracy of the non-defective product determination by the new inspection device. It becomes possible to confirm.

Therefore, the non-defective product determination standard setting device of the above-described inspection apparatus described in the present application further includes a non-defective product determination standard set by the above-described method of the present invention (for example, M _NG ± kσ _NG (K is a positive number, preferably k = 3, 4, 5 or 6, more preferably k = 6), and is set in the new testing device from the beginning. A comparison determination processing unit that compares the non-defective product determination standard can be provided.

  This can also be employed when the inspection apparatus is the image inspection apparatus described above.

  According to the non-defective product determination standard setting method of the inspection apparatus (for example, image inspection apparatus) described in the present application and the non-defective product determination standard setting apparatus of the inspection apparatus (for example, image inspection apparatus) described above, An inspection device that determines the quality of an inspection object based on an inspection measurement value obtained by using the inspection apparatus, for example, an inspection based on an inspection measurement value obtained from an image signal captured by an imaging unit that captures an image of the inspection object A non-defective product criterion setting method for setting a non-defective product criterion for determining whether a test object is a non-defective product or a non-defective product in an image inspection apparatus that determines the quality of an object. Non-defective product judgment ”, that is, a non-defective product judgment standard setting method and an inspection device that can achieve the purpose of quality control of eliminating“ defective product judgment ”(= statistically zero) It is possible to provide a definitive quality criterion setting device.

Further, the present patent application describes a determination processing accuracy determination method of the inspection apparatus. This is a method for determining the accuracy of the determination process in an inspection apparatus for determining the quality of an inspection object based on an inspection measurement value obtained using the inspection apparatus for the inspection object, and a plurality of non-defective products, Assuming that a plurality of non-defective product weight values obtained by using the inspection device are represented by a normal distribution, a non-defective product average value _MG and a non-defective product standard deviation σ _G are obtained, Assuming that a plurality of inspection metric values obtained by using the apparatus are represented by a normal distribution, an inspection average value M _S and an inspection standard deviation σ _S are obtained, and predetermined with the non-defective product average value _MG as a center. a good average range which is set in a range that is, comparing the test mean value M _S, or non-defective standard deviation is set in a range which is predetermined about said non-defective standard deviation sigma _G Range and the inspection standard deviation σ _S Comparison, or a good average range set in the range is predetermined about said good average value M _G, about said non-defective standard deviation sigma _G with comparing the test mean value M _S The non-defective product standard deviation range set in a predetermined range is compared with the inspection standard deviation σ _S.

The determination processing accuracy determination device that determines the accuracy of the determination processing in the inspection device that executes the method is an inspection that determines the quality of the inspection object based on the inspection measurement value obtained by using the inspection device for the inspection object. It is a determination processing accuracy determination device that determines the accuracy of the determination processing in the device, and for a plurality of non-defective products, a plurality of non-defective product weight values obtained using the inspection device are assumed to be represented by a normal distribution. the average value M _G, a first non-defective mean, standard deviation calculation means for calculating a non-defective standard deviation sigma _G, the plurality of test object, a plurality of test weighings obtained using the inspection system regular assuming represented by distribution, predetermined as the center and the examination average value M _S, a first inspection mean, standard deviation calculation means for calculating an inspection standard deviation sigma _S, the good average value M _G Set to the specified range. And a good average range is, the first good average and inspection average value comparison processing means for comparing the test mean value M _S, or set within a range which is determined in advance around the non-defective standard deviation sigma _G and good standard deviation range that is, the first non-defective standard deviation and inspection standard deviation comparison processing means for comparing the test standard deviation sigma _S, or ranges are predetermined around the good average value M _G a good average range that is set in a first non-defective average and inspection average value comparison processing means for comparing the test mean value M _S, a range which is predetermined about a good standard deviation sigma _G Both the set non-defective product standard deviation range and the first non-defective product standard deviation / inspection standard deviation comparison processing means for comparing the inspection standard deviation σ _S are provided.

  The determination processing accuracy determination method of the inspection apparatus and the determination processing accuracy determination apparatus of the inspection apparatus can also be applied when the inspection apparatus is the image inspection apparatus described above.

That is, the determination processing accuracy determination method of the image inspection apparatus described in the present patent application determines the quality of the inspection object based on the inspection measurement value obtained from the image signal captured by the imaging unit that captures the image of the inspection object. In the image inspection apparatus for determining, a method for determining the accuracy of the determination processing, wherein a plurality of non-defective product weight values obtained from an image signal captured by the imaging means are used for a plurality of non-defective products. Assuming that it is expressed by a distribution, a non-defective product average value _MG and a non-defective product standard deviation σ _G are obtained, and a plurality of inspection objects are obtained from the image signal captured by the imaging means using the image inspection apparatus. and a plurality of test weight value is assumed to be represented by a normal distribution, and the inspection average M _S, obtains an inspection standard deviation sigma _S, predefined about said good average value M _G A good average range set in the range that, comparing the test mean value M _S, or a non-defective standard deviation range set in the range is predetermined about said non-defective standard deviation sigma _G , comparing the test standard deviation sigma _S, or a good average range set in the range is predetermined about said good average value M _G, with comparing the test mean value M _S The non-defective product standard deviation range set in a predetermined range centering on the non-defective product standard deviation σ _G is compared with the inspection standard deviation σ _S.

The determination processing accuracy determination apparatus for determining the accuracy of the determination process in the image inspection apparatus that executes the method is an inspection object based on an inspection measurement value obtained from an image signal captured by an imaging unit that captures an image of the inspection object. Measuring the accuracy of the determination process in an image inspection apparatus for determining the quality of an object, and measuring a plurality of non-defective products obtained from an image signal captured by the imaging means using the image inspection apparatus. assuming the value is represented by a normal distribution, a good average value M _G, and a second good average, standard deviation calculation means for calculating a non-defective standard deviation sigma _G, a plurality of the inspection object, the Assuming that a plurality of inspection metric values obtained from the image signal captured by the imaging means using an image inspection apparatus are represented by a normal distribution, the inspection average value M _S and the inspection standard deviation σ A second test average value, standard deviation calculation means for calculating the _S, a good average range set in the range is predetermined about said good average value M _G, and the test mean value M _S Second non-defective product average value / inspection average value comparison processing means, or non-defective product standard deviation range set in a predetermined range centering on the non-defective product standard deviation σ _G , and the inspection standard deviation σ second good standard deviation and inspection standard deviation comparison processing means for comparing the _S, or a good average range set in the range is predetermined about said good average value M _G, the test mean value a second good average and inspection average value comparison processing means for comparing the M _S, and the non-defective standard deviation range is set in a range which is predetermined about a good standard deviation sigma _G, the inspection standard deviation and σ _S Both the second non-defective standard deviation and inspection standard deviation comparison processing means for compare, and is characterized in that it comprises a.

Determination method accuracy determination method for such inspection apparatus, determination processing accuracy determination apparatus for determining accuracy of determination processing in inspection apparatus, determination processing accuracy determination method for image inspection apparatus, and determination processing accuracy determination in image inspection apparatus According to the determination processing accuracy determination device for a good average range which is set in a range which is predetermined about a good average value M _G, by comparing the test mean value M _S, inspection weighing The drift of the value can be grasped. Further, by comparing the non-defective product standard deviation range set in a predetermined range centering on the non-defective product standard deviation σ _G with the inspection standard deviation σ _S , it is possible to grasp the variation of the inspection measurement value. it can.

  Then, by grasping the drift of the inspection measurement value and the variation of the inspection measurement value, the variation of the judgment accuracy caused by the disturbance such as the dirt of the lens of the CCD camera during the inspection process is grasped. It is possible to determine whether or not the set determination system is maintained.

  According to the present invention, an inspection apparatus that determines the quality of an inspection object based on an inspection measurement value obtained by using the inspection apparatus for the inspection object, for example, an image signal captured by an imaging unit that captures an image of the inspection object. In the image inspection apparatus that determines the quality of the inspection object based on the inspection measurement value obtained from the non-defective product determination standard setting method for setting the non-defective product determination standard for determining whether the inspection object is a non-defective product or a defective product Therefore, it is possible to achieve the quality control purpose of “determining a defective product as a non-defective product”, that is, eliminating “defective product determination” (= statistically zero). It is possible to provide a non-defective product determination criterion setting method and a quality determination criterion setting device in an inspection apparatus.

The figure showing distribution of the measured value by the image inspection apparatus about a some non-defective product (1 spoon) and a some defective product (0 spoon, 2). The figure explaining the state which converts the non-defective product distribution state represented by a curve into a straight line display. A process for setting a non-defective product determination criterion will be described with reference to a distribution state displayed in a straight line on the screen. The figure explaining the process of determining the dispersion | distribution of an average value with reference to the distribution state of the test | inspection average value currently displayed on the screen with the straight line. The side view explaining the imaging means part of an image inspection apparatus. The figure explaining the method of setting non-defective item judgment standard setting based only on non-defective item distribution. The figure explaining in more detail the method of setting a non-defective product criterion setting based only on a non-defective product distribution. The apparatus block schematic diagram explaining the apparatus block of this invention.

  Hereinafter, with reference to the accompanying drawings, a non-defective product judgment criterion setting method and a quality judgment standard setting device in an inspection device, and a method for judging the accuracy of a quality judgment process and a judgment processing accuracy judgment device in an inspection device are applied to an image inspection device. The case where it will be described.

  FIG. 5 illustrates an example of an imaging unit in the image inspection apparatus 1 to which the quality determination criterion setting device 2 and the determination processing accuracy determination device 3 are applied.

  When the inspection object 100 is conveyed by the conveyance conveyor 101, an image of the inspection object 100 is captured by the CCD camera 102 provided on the upper side of the conveyance conveyor 101. In the example of FIG. 5, the inspection object 100 is a cylindrical milk can having a diameter of 370 mm, the milk can is filled with powdered milk, and the state of the spoon placed on the powdered milk is checked by an image inspection apparatus. Inspected.

  In FIG. 5, what are indicated by reference numerals 103 and 104 are a pinhole lens and a near-infrared LED, respectively. The illumination light is uniformly applied to the inside of the inspection object 100 (cylindrical milk can having a diameter of 370 mm) by the ring-shaped illumination 105 and the diffusion plate 106 disposed at a height of 30 mm from the upper end of the inspection object 100. ing.

  In the example shown in FIG. 5, as the state of the spoon placed on the powdered milk, the number of spoons (one that is the exact number is placed or one is not placed by mistake = 0, In order to confirm whether or not there are a plurality of such objects, the near red color represented by reference numeral 104 is used so that the target is binarized with near-infrared light and inspected by an image inspection apparatus. Outside LED is used. If you want to check the state of the spoon placed on top of the powdered milk, whether the color of the spoon (blue, yellow, etc., is a color corresponding to the type of powdered milk loaded in the milk can) , White light is used for the portion indicated by reference numeral 104, and inspection can be performed by performing a vector analysis on the three elements (luminance, hue, color) of the color with an image inspection apparatus.

  The image of the inspection object 100 captured by the CCD camera 102 is sent to the non-defective product determination reference setting device 2 or the determination processing accuracy determination device 3 via the image inspection device 1 as shown in FIG.

  The image inspection apparatus 1, the non-defective product determination reference setting apparatus 2, and the determination processing accuracy determination apparatus 3 are all information acquisition means including an interface unit, a keyboard, a pointing device, etc. that enable information exchange and bidirectional communication with the outside, CRT A computer program comprising an information output means, a storage unit, an arithmetic processing unit, etc., comprising an image display unit on a screen such as a screen, a printing unit and the interface unit, and causing the computer to execute each processing operation described below. It can be configured by a computer such as a personal computer that performs each processing operation described below in accordance with the command and the command and information acquired by the information acquisition means as described above.

  Hereinafter, as the state of the spoon placed on the powdered milk loaded inside the inspection object 100 (cylindrical milk can with a diameter of 370 mm), the number of spoons (whether one is the exact number is placed) In the processing operation for confirming whether or not one or more are mistakenly placed, or the like, the non-defective product judgment criterion setting device 2 and the judgment processing accuracy judgment device 3 of the present invention are confirmed. The non-defective product judgment reference setting method executed by the method and the judgment processing accuracy judgment method of the present invention will be described.

In the non-defective product determination standard setting method, first, the number of spoons placed on a plurality of non-defective products (for example, powdered milk loaded in a cylindrical milk can which is the test object 100) is accurate as the test object 100. one) with, on the assumption that a plurality of non-defective weighed values obtained from the image signal captured by the CCD camera 102 of the image inspection device 1 is represented by a normal distribution, a good average value M _G is a number, A process for obtaining a good standard deviation σ _G is performed.

This process is performed by non-defective mean, standard deviation computing unit 21 is a processing operation of the computer in good criterion setting unit 2, a good average value M _G obtained, good standard deviation sigma _G, good distribution It is stored as data in the storage unit.

The distribution of the non-defective products obtained in this way is displayed on the CRT screen 25 of the non-defective product determination criterion setting device 2 by the image display unit 24 of the non-defective product determination criterion setting device 2 or the CRT screen 12 of the image inspection device 1 by the image display unit 11 of the image inspection device 1. When the image is displayed, for example, a normal distribution curve represented by reference numeral 40 in FIGS. 1 and 2 is obtained. Note the normal distribution curve, the image display unit 24, the image display unit 11, a good average value M _G, and displays an image with different colors for each region where an integral multiple of the non-defective standard deviation sigma _G is added or subtracted be able to.

Next, in the non-defective product determination criterion setting method, the number of spoons placed on a plurality of defective products as the inspection object 100 (for example, powdered milk loaded in a cylindrical milk can which is the inspection object 100). Is an inaccurate number of 0 (regular number −1) or plural numbers (in this example, 2 (regular number +1)), and the image inspection apparatus 1 uses the image signal captured by the CCD camera 102. Assuming that the obtained plurality of defective product measured values are represented by a normal distribution, a process for obtaining a defective product average value M _NG and a defective product standard deviation σ _NG is performed.

This processing is performed by the defective product average value / standard deviation calculating means 22 which is a processing operation unit of the computer in the non-defective product determination standard setting device 2, and the obtained defective product average value M _NG and the defective product standard deviation σ _NG , The defective product distribution is stored as data in the storage unit.

The distribution of defective products obtained in this way is displayed on the CRT screen 25 of the non-defective product criterion setting device 2 by the image display device 24 of the non-defective product criterion setting device 2 or the CRT screen of the image inspection device 1 by the image display device 11 of the image inspection device 1. When the image is displayed on 12, for example, the normal distribution curves represented by reference numerals 41 and 42 in FIG. 1 are obtained (the number of spoons in which the normal distribution curve represented by reference numeral 41 is placed on the powdered milk is 0. This is a case where the number of spoons in which the normal distribution curve represented by the book 42 is placed on the powdered milk is 2). The normal distribution curve is also displayed by the image display means 24 and the image display means 11 in different colors for each region where an integral multiple of the defective standard deviation σ _NG is added or subtracted from the defective product average value M _NG. can do.

Both of the image display means 24 of the non-defective product criterion setting apparatus 2 and the image display means 11 of the image inspection apparatus 1 that are processing operation units of a computer express the normal distribution curves 40, 41, and 42 as described above with straight lines. For example, the normal distribution curve 40 shown in FIG. 2 can be displayed on a straight line represented by reference numeral 43 in FIG. May be displayed in this manner in a straight line, the image display unit 24, the image display unit 11, a good average value M _G, each area where an integral multiple of the non-defective standard deviation sigma _G is added or subtracted, defective average M from _NG, it is possible to display images with different colors for each region where an integral multiple of the defective standard deviation sigma _NG is added or subtracted.

  The non-defective product determination reference setting device 2 includes a distribution state image display means 23 which is a processing operation unit of a computer. By the distribution state image display means 23, the non-defective product measurement value stored in the storage unit as described above. And the distribution of defective product weighing values can be compared and displayed on the same screen.

  For example, the distribution state image display means 23, as shown in FIG. 1, shows the distribution of non-defective product measurement values and the distribution of defective product measurement values using the normal distribution curves 40, 41, 42 described above in the normal distribution curve state. Compare and display on the same screen. Alternatively, as shown in FIG. 3, the distribution of non-defective product measurement values and the distribution of defective product measurement values are compared and displayed on the same screen in a straight line state.

  In this way, the step of setting the non-defective product determination standard by comparing the non-defective product measurement value distribution with the defective product measurement value distribution in the non-defective product determination standard setting method is performed.

For example, as shown in FIG. 1, in the distribution of non-defective product weight values (normal distribution curve 40), a position obtained by adding 6σ _NG to M _NG that is the average value of defective products in the distribution of defective product weight values (normal distribution curve 41). Can be set as the upper limit non-defective product determination criterion, and the position obtained by subtracting 6σ _NG from M _NG, which is the average value of defective products in the distribution of defective product measurement values (normal distribution curve 42).

According to the experiments by the inventors, in the distribution of the non-defective product measurement value (normal distribution curve 40 in FIG. 1), the defective product average value M _NG + 6σ of the distribution of defective product measurement values (normal distribution curve 41 in FIG. 1). _{If NG is set} to the lower limit non-defective product judgment criterion and M _NG -6σ _NG , which is the average value of defective products in the distribution of defective product measurement values (normal distribution curve 42 in FIG. 1), is set to the upper limit non-defective product judgment criterion, It was possible to eradicate “determining that the product is non-defective”, that is, eliminating “determining non-defective product” (= statistically zero).

According to the method described above, in the method shown in FIGS. 6 and 7, that is, in the distribution of the non-defective product weight values (for example, the normal distribution curve 40 in FIG. 1), the non-defective product weight value distribution (the normal value in FIG. the lower limit of the non-defective criteria (integer multiple of M _G - [sigma] _G) by adding or subtracting the integer multiple of the non-defective standard deviation sigma _G against a good average value in the distribution curve 40) M _G, upper Compared with the method of setting the non-defective product judgment standard (positive multiple of M _G + σ _G ), the number of inspection objects that are judged as “defective products” increases despite being good products. It becomes possible to eradicate “determining a defective product as a non-defective product”, that is, to eliminate “determining a non-defective product as good” (= statistically zero).

In the above description, the distribution of non-defective product measurement values (normal distribution curve 41) in the non-defective product measurement value distribution (normal distribution curve 40) is determined by the determination standard setting processing means 27 which is a processing operation unit of the computer in the non-defective product determination reference setting device 2. M _NG + kσ _NG (k is a positive number), which is the average value of defective products, and M _NG −kσ _NG (k, which is the average value of defective products in the distribution of defective product measurement values (normal distribution curve 42). Is set to the upper limit non-defective product criterion, as shown in FIG. 1 or 2, the distribution of non-defective product weight values displayed on the CRT screen 25 of the non-defective product criterion setting device 2 Based on the comparison display with the distribution of the non-defective product measurement value, the person operating the non-defective product determination standard setting device 2 displays the distribution of the defective product measurement value on the screen from the distribution of the non-defective product measurement value (normal distribution curve 40). Above (normal distribution curve 41) When, it is also possible to perform the processing for setting the position corresponding to the lower limit of the distribution of defective weighing (normal distribution curve 42) good criterion lower limit, respectively, the non-defective criteria limit.

As described above, the image display unit 24, the image display unit 11, a good average value M _G, each area where an integral multiple of the non-defective standard deviation sigma _G is added or subtracted, from the defective average M _NG, defective Since it is possible to display an image with different colors for each region where an integral multiple of the standard deviation σ _NG is added or subtracted, the person who operates the non-defective product determination criterion setting device 2 can display the lower limit non-defective product on the screen. Judgment criteria and upper limit non-defective product judgment criteria can be easily set.

  FIG. 3 shows a normal distribution curve 40 of a non-defective product (in the case of the normal distribution curve represented by reference numeral 40 in FIG. 1 and the number of spoons placed on the powdered milk is a normal one) in a straight line shape. A line segment 45 representing the normal distribution curve of the line segment 43 and the defective product (in the case where the number of spoons placed on the powdered milk is a normal number-1 = 1) 45 in a straight line. And the line segment 44 representing the normal distribution curve of the defective product (when the number of spoons placed on the powdered milk is a normal number + 1 + 1 = 2) is displayed on the screen. It explains the state of being. The processing operation unit of the non-defective product determination criterion setting device 2 compares the lower limit of the line segment 43 and the upper limit of the line segment 45 to set the lower limit non-defective product determination criterion, and sets the upper limit of the line segment 43 and the lower limit of the line segment 44 to the computer. It is also possible to set the upper limit non-defective product judgment standard by the processing operation unit of the above, or the person who operates the non-defective product judgment standard setting device 2 on the screen, A non-defective product judgment standard can also be set.

In addition, the non-defective product determination criterion setting device 2 described above is further provided with a determination processing accuracy comparison / determination unit 26 including a processing operation unit of a computer. The lower limit non-defective product judgment criteria (M _NG + kσ _NG (k is a positive number) that is the average value of defective products in the distribution of defective product measurement values (normal distribution curve 41)) and / or the upper non-defective product judgment criteria (defective products) _NG- kσ _NG (k is a positive number), which is the average value of defective products in the distribution of the measured values (normal distribution curve 42), is determined from the beginning for the image inspection apparatus to be newly used. Compared with (generally a multiple of M _G ± σ _G , which is an average value of non-defective products), it is possible to perform processing for confirming the accuracy of non-defective product determination by the image inspection apparatus that is newly used. .

  That is, before actually starting the inspection of the inspection object 100 using the new image inspection apparatus 1, what is the non-defective product determination criterion set in the new image inspection apparatus 1 from the beginning? The accuracy of the non-defective product determination by the new image inspection apparatus 1 can be confirmed by the non-defective product determination standard setting device 2, and this process is performed according to the above-described non-defective product determination standard setting method of the inspection device and non-defective product of the image inspection device. It can be added to the criteria setting method.

  As described above, according to the non-defective product determination reference setting method in the image inspection apparatus 1 and the pass / fail determination reference setting apparatus 2 in the image inspection apparatus 1, the inspection measurement obtained from the image signal captured by the CCD camera 102 that captures the image of the inspection object 100. In the image inspection apparatus 1 that determines the quality of the inspection object 100 based on the value, “determines that a defective product is a non-defective product”, that is, “determines a non-defective product” (= statistical). It is possible to set a non-defective product determination criterion (a non-defective product determination criterion for determining whether the inspection object 100 is a non-defective product or a non-defective product) that makes it possible to achieve the quality control purpose of setting it to zero.

  Next, the determination processing accuracy determination method for the image inspection apparatus 1 according to the present invention and the determination processing accuracy determination apparatus 3 for determining the accuracy of the determination processing in the image inspection apparatus 1 will be described.

  This determination processing accuracy determination method and determination processing accuracy determination device 3 can perform self-diagnosis in the course of performing an inspection as to whether or not the above-described image inspection device 1 continues to maintain the accuracy of the non-defective product determination criteria once set. It is to make.

  For example, in the process of continuing the inspection, the inspection object 100 is being transported due to a decrease in the amount of light due to deterioration of the light source of the CCD camera 102 over time, contamination of the lens, attachment of the CCD camera 102, etc. If the determination accuracy fluctuates due to knocking of the conveyer 101, misalignment of guides, noise of communication cables connecting various devices, etc., the determination accuracy of the image inspection apparatus 1 may fluctuate due to these disturbances. However, it is possible to make a self-diagnosis in the course of the inspection whether or not the determination accuracy initially set in the image inspection apparatus 1 is maintained.

In the determination processing accuracy determination method in the image inspection apparatus 1 of the present invention, first, a plurality of non-defective products (for example, powdered milk loaded in a cylindrical milk can that is the inspection object 100) as the inspection object 100. Assuming that the number of non-defective products obtained from the image signal captured by the CCD camera 102 of the image inspection apparatus 1 is represented by a normal distribution using the exact number of spoons placed. Thus, a process for obtaining the non-defective product average value _MG and the non-defective product standard deviation σ _G is performed.

This process is performed by non-defective mean, standard deviation calculation means 31 is a processing operation of the computer in the determination processing precision determination unit 3, and a good average value M _G obtained, good standard deviation sigma _G, good distribution It is stored in the storage unit as data.

The determination processing accuracy determination device 3 is further provided with a non-defective product average range setting means 32 which is a processing operation unit of a computer. This good average range setting means 32, based on the data stored in the storage unit of the determination processing accuracy determination unit 3, good around the good average value M _G obtained by the good average, standard deviation calculation means 31 The average range is set to a predetermined range. The non-defective product average range thus set is also stored in the storage unit of the determination processing accuracy determination device 3.

The determination processing accuracy determination apparatus 3 further includes a non-defective standard deviation range setting unit 33 that is a processing operation unit of a computer. Based on the data stored in the storage unit of the determination processing accuracy determination device 3 by the non-defective product standard deviation range setting unit 33, the non-defective product standard deviation σ _G obtained by the non-defective product average value / standard deviation calculating unit 31 is used as a center. The non-defective product standard deviation range is set to a predetermined range. The non-defective standard deviation range thus set is also stored in the storage unit of the determination processing accuracy determination device 3.

Further, the determination processing accuracy determination device 3 includes an inspection average value / standard deviation calculation means 34 which is a processing operation unit of a computer. The inspection average value / standard deviation calculating means 34 represents a plurality of inspection metric values obtained from the image signal captured by the CCD camera 102 in a normal distribution when the image inspection apparatus 1 inspects a plurality of inspection objects 100. Assuming that the inspection average value M _S and the inspection standard deviation σ _S are calculated, the processing is performed.

The inspection average value M _S , the inspection standard deviation σ _S , and the inspection metric value distribution obtained by the calculation process of the determination processing accuracy determination device 3 are stored in the storage unit of the determination processing accuracy determination device 3, respectively.

The determination processing accuracy determination device 3 includes an image display means 35, and the distribution of the inspection object for which the inspection average value M _S and the inspection standard deviation σ _S are obtained as described above is a normal distribution as shown in FIG. It can be displayed on the CRT screen 36 of the determination processing accuracy determination device 3 like a curve or a line segment 43 shown in FIG.

The determination processing accuracy determination device 3 includes non-defective product average value / inspection average value comparison processing means 37 which is a processing operation unit of a computer. The by good average and inspection average value comparison processing unit 37, a non-defective average range which has been set by the good average range setting means 32 as described above, the processing of the test average value M _S is compared is performed.

Further, the determination processing accuracy determination device 3 includes a non-defective product standard deviation / inspection standard deviation comparison processing means 38 which is a processing operation unit of a computer. The non-defective product standard deviation / inspection standard deviation comparison processing unit 38 performs processing for comparing the non-defective product standard deviation range set by the non-defective product standard deviation range setting unit 33 with the inspection standard deviation σ _{S as} described above. Is called.

FIG. 4 shows an image of the distribution of the inspection object whose inspection average value M _S and inspection standard deviation σ _S are obtained by the inspection average value / standard deviation calculating means 34 on the CRT screen 36 of the determination processing accuracy determination device 3. Although it is displayed as a straight line segment by means 35, depending on whether the lower limit value and the upper limit value of this line segment are within the non-defective product average range, the non-defective product average value / inspection average value comparison processing means 37 A process of comparing the inspection average value M _S and the inspection average value M _S can be performed.

Similarly, the non-defective product standard deviation / inspection standard deviation comparison processing means 38 performs a process of comparing the non-defective product standard deviation range set by the non-defective product standard deviation range setting means 33 with the inspection standard deviation σ _S. it can.

The good average and inspection average value comparison processing unit 37, by comparing the non-defective average range, and the inspection average M _S, it is possible to grasp the drift of the test metric. Further, the non-defective product standard deviation / inspection standard deviation comparison processing means 38 compares the non-defective product standard deviation range with the inspection standard deviation σ _S so that the variation in the inspection measurement value can be grasped. Therefore, it is possible to adopt a form in which only one of the good product average value / inspection average value comparison processing means 37 and the good product standard deviation / inspection standard deviation comparison processing means 38 is provided, or a form in which both are provided. It can also be.

  By executing the above-described determination processing accuracy determination method of the image inspection apparatus 1 by the determination processing accuracy determination apparatus 3 that determines the accuracy of the determination process in the image inspection apparatus 1 as described above, Grasping variation and grasping fluctuations of judgment accuracy caused by disturbances such as dirt on the lens of the CCD camera during the inspection process, and judging whether the judgment system set at the initial stage is maintained it can.

The above-described determination processing accuracy determination device 3 is further provided with a warning information generation processing unit 39 including a processing operation unit of a computer, and the non-defective product average value / inspection average value comparison processing means 37 or the non-defective product. standard deviation and inspection standard deviation comparison processing unit 38, or the result of the determination processing by these both test average value M _S deviates a good average range, or, departing from the non-defective standard deviation range check standard deviation sigma _S In the case where the warning information generation processing unit 39 performs a processing operation, a warning is output as image information on the CRT screen of the determination processing accuracy determination device 3, or the determination processing accuracy determination device 3 The warning information can be output as voice information from the information output unit.

  In this way, if the determination accuracy initially set for the image inspection apparatus 1 is not maintained, it is immediately grasped and necessary countermeasures can be taken.

  Further, the non-defective product average value / inspection average value comparison processing unit 37 and the non-defective product standard deviation / inspection standard deviation comparison processing unit 38 are configured to perform the comparison for a predetermined number of times in a predetermined period, for example, once per operation day. On the other hand, the determination processing accuracy determination device 3 is further provided with a determination information recording unit 30 including a processing operation unit of a computer so that the non-defective product average value / inspection average value is obtained. The result of the comparison determination process performed every day by the comparison processing unit 37 and the non-defective product standard deviation / inspection standard deviation comparison processing unit 38 is recorded in the storage unit of the determination processing accuracy determination device 3 as a record (daily report). Can do.

  If this record (daily report) recorded in the storage unit of the determination processing accuracy determination device 3 can be printed and taken out from the information output means of the determination processing accuracy determination device 3 as required, the image inspection apparatus It can be confirmed later whether or not the inspection has been continuously performed with the determination accuracy set to 1 at the initial stage.

  In the above description, the case where the image inspection device 1, the non-defective product determination reference setting device 2, and the determination processing accuracy determination device 3 are each configured by separate computers has been described. However, the image inspection device 1 including a computer is described above. The non-defective product average value / standard deviation calculating means 21, the defective product average value / standard deviation calculating means 22, the distribution state image display means 23, the image display means 24, the determination processing accuracy comparison / determination means 26, and the determination reference setting are performed. Processing means 27, determination information recording unit 30, good product average value / standard deviation calculation means 31, good product average range setting means 32, good product standard deviation range setting means 33, inspection average value / standard deviation calculation means 34, good product average value / inspection The average value comparison processing means 37, the non-defective product standard deviation / inspection standard deviation comparison processing means 38, and the warning information generation processing unit 39 are provided, and the above-mentioned non-defective product determination is performed. Quasi setting, determination processing precision determination method may be to the image inspection apparatus also performs.

  The preferred embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to such embodiments. For example, by inspecting the weight of an object to be inspected, an inspection for determining “a product to be inspected within the proper weight range” (good product) and “a product to be inspected outside the proper weight range” (defective product) Inspection equipment that determines "inspection product within proper size" (non-defective product) and "inspection product outside proper size" (defective product) by inspecting the size of the device and inspection object The present invention can be applied to various inspection apparatuses.

DESCRIPTION OF SYMBOLS 1 Image inspection apparatus 11 Image display means 12 CRT screen 2 Non-defective product determination standard setting device 21 Non-defective product average value / standard deviation calculation means 22 Defective product average value / standard deviation calculation means 23 Distribution state image display means 24 Image display means 25 CRT screen 26 Determination processing accuracy comparison determination means 27 Determination standard setting processing means 3 Determination processing accuracy determination device 31 Non-defective product average value / standard deviation calculating means 32 Non-defective product average range setting means 33 Non-defective product standard deviation range setting means 34 Inspection average value / standard deviation calculating means 35 Image display means 36 CRT screen 37 Non-defective product average value / inspection average value comparison processing means 38 Non-defective product standard deviation / inspection standard deviation comparison processing means 39 Warning information generation processing section 30 Determination information recording section 40 Normal distribution curve of non-defective products (the number of spoons is 1)
41 Normal distribution curve of defective products (when the number of spoons is 0)
42 Normal distribution curve of defective products (when the number of spoons is 2)
43 Line segments 44 and 45 showing normal distribution curve 40 of non-defective product in a straight line 100 Line segment 100 expressing normal distribution curve of defective product in a straight line 100 Inspection object 101 Conveyor 102 CCD camera 103 Pinhole lens 104 Near infrared LED
105 Ring illumination 106 Diffuser

Claims (3)

In an inspection apparatus that determines whether the inspection object is good or not using the packaging container of the food filled with food as the inspection object, it is determined whether the size of the inspection object in the inspection object is within an appropriate size. A non-defective product criterion setting method for setting a non-defective product criterion.
The food container filled with the food is a milk can filled with powdered milk,
The inspection apparatus determines whether or not the size of the inspection object in the inspection object is within an appropriate size based on the inspection measurement value obtained from the image signal captured by the imaging unit that captures the image of the inspection object. An image inspection apparatus that performs a binarization process using near-infrared light on the inspection target for determining whether or not the inspection target is within an appropriate size, so that the size of the inspection target is within an appropriate size And inspecting whether
For a plurality of non-defective products whose size to be inspected is within an appropriate size, a distribution of non-defective product weight values obtained using the inspection device and a plurality of defective products whose size to be inspected is outside the proper size Obtain the distribution of measured values of defective products obtained using the inspection device,
The distribution of the non-defective product measurement values and the distribution of the defective product measurement values are displayed on the same screen with the number of detected pixels on the horizontal axis and the number of samples on the vertical axis, and the two are compared on the same screen. ,
The upper non-defective product criterion is set to be smaller than the lower limit of the non-defective product weight distribution on the upper limit side in the non-defective product weight distribution,
A non-defective product judgment criterion setting method for setting a non-defective product judgment criterion by setting a lower acceptable product criteria to a value larger than the upper limit of the distribution of defective product measurement values on the lower limit side. The size of the inspection object in the inspection object is based on the inspection measurement value obtained from the image signal captured by the imaging means that captures the image of the inspection object using the packaging container of the food filled with food as the inspection object. A non-defective product criterion setting method for setting a non-defective product criterion for determining whether or not the size of an inspection object in the inspection object is within an appropriate size in an image inspection apparatus that determines whether or not the image is within an appropriate size Because
The food container filled with the food is a milk can filled with powdered milk,
Whether the size of the inspection object in the inspection object is within the appropriate size by performing binarization processing by near infrared rays on the inspection object for determining whether the inspection object is within the appropriate size or not. And a plurality of non-defective product weight values obtained from the image signal captured by the image pickup means using the image inspection apparatus are normal. Assuming that it is expressed by a distribution, a non-defective product average value _MG and a non-defective product standard deviation σ _G are obtained,
For a plurality of defective products in which the size of the inspection target in the inspection target object is outside the proper size, the plurality of defective product measurement values obtained from the image signal captured by the imaging unit are in a normal distribution using the image inspection apparatus. Assuming that the average value M _NG of defective products and the standard deviation σ _{NG of} defective products are obtained,
The distribution of the non-defective product measurement value and the distribution of the defective product measurement value are displayed on the same screen with the horizontal axis indicating the number of detected pixels and the vertical axis indicating the number of samples.
Comparing the distribution of non-defective product measurement values with the distribution of non-defective product measurement values on the same screen, in the distribution of non-defective product measurement values, the inspection measurement value corresponding to M _NG ± kσ _NG (k = 6) is determined as non-defective product By setting the standard, the upper non-defective product judgment standard is set to be smaller than the lower limit of the non-defective product weight distribution on the upper limit side in the non-defective product weight distribution.
A non-defective product criterion setting method for an image inspection apparatus, characterized in that the lower limit non-defective product criterion is set to be larger than the upper limit of the distribution of defective product measurement values on the lower limit side. The size of the inspection object in the inspection object is based on the inspection measurement value obtained from the image signal captured by the imaging means that captures the image of the inspection object using the packaging container of the food filled with food as the inspection object. A non-defective product determination criterion setting device for setting a non-defective product determination criterion for determining whether or not the size of the inspection target in the inspection object in the image inspection device for determining whether or not the proper size is within the proper size. There,
The food container filled with the food is a milk can filled with powdered milk,
The image inspection apparatus performs a binarization process using near infrared rays on the inspection object for determining whether or not the inspection object is within an appropriate size, whereby the inspection object size in the inspection object is within the appropriate size. And inspecting
For a plurality of non-defective products in which the size of the inspection target in the inspection object is within an appropriate size, a plurality of non-defective product weight values obtained from the image signal captured by the imaging unit using the image inspection device are represented by a normal distribution. assuming that a good average value M _G, and good products mean, standard deviation calculation means you calculating a good standard deviation sigma _G,
For a plurality of defective products in which the size of the inspection target in the inspection target object is outside the proper size, the plurality of defective product measurement values obtained from the image signal captured by the imaging unit are in a normal distribution using the image inspection apparatus. assuming represented, and defective average M _NG, a defective mean, standard deviation calculation means you calculating a defective standard deviation sigma _NG,
The distribution of the non-defective metering value, and the distribution of the defective weighing, the number of detection pixels on the horizontal axis, vertical axis and displayed on the same screen as the number of samples, M _G ± k? Of the distribution of the non-defective weighing _{G (k} = 6) and the range of the defective distribution of metric values _{M NG ± kσ NG (k =} 6) in the range and the same that displays compared on the screen the distribution state image display means By setting the upper and lower non-defective product judgment criteria to be smaller than the lower limit of the non-defective product weight value distribution on the upper limit side in the non-defective product weight value distribution,
A non-defective product criterion setting device for an image inspection apparatus, characterized in that the lower limit non-defective product criterion is set to be larger than the upper limit of the distribution of defective product measurement values on the lower limit side.

Images