This application is based on and claims priority under 35 USC 119 from Japanese Patent Application No. 2021-151906 filed Sep. 17, 2021.
The present invention relates to a collation device, and a non-transitory computer readable medium storing a collation program.
JP2018-026147A describes an object management device that, in a case of performing individual management based on a microscopic pattern on an object surface, observes a microscopic pattern from an identical image area including a posture, at the time of registration and at the time of collation. Specifically, JP2018-026147A describes the object management device that manages an object based on a microscopic pattern on a surface of an object included in an image obtained by imaging the surface of the object. The object management device has a first position correction unit that performs registration of captured images based on a microscopic pattern commonly appearing on a plurality of objects. The object management device has a second position correction unit that performs registration of the captured images based on a non-microscopic pattern commonly appearing on a plurality of objects and having a size greater than the microscopic pattern before the registration by the first position correction unit.
JP2017-058306A describes an identification device including an imaging unit for registration that images two facing side surfaces of a registration target substrate from a direction perpendicular to a transport direction, and a substrate identification unit that extracts a feature quantity from image data of the registration target substrate captured by the imaging unit for registration, stores the extracted feature quantity of the registration target substrate in association with substrate information, and identifies a collation target substrate by collating the feature quantity extracted from image data of a collation target substrate with the stored feature quantity of the registration target substrate.
JP2017-183390A describes an identification device including an imaging unit that images at least three places including a front side surface and a rear side surface of a target substrate with a plurality of cameras imaging a side surface of the target substrate being transported in a manufacturing line from any one of the front or the rear with respect to a transport direction, and a substrate identification unit that extracts a feature quantity from image data of the target substrate captured by the imaging unit, registers the target substrate while associating the extracted feature quantity with substrate information of the target substrate, and collates the feature quantity extracted from the image data of the target substrate with the registered feature quantity to identify the target substrate.
In a case of imaging a random pattern depending on a microscopic pattern of an object and collating a captured image with a registration image registered by imaging an identical position in advance to uniquely identify the object, in a case where the object has an elliptic cylindrical shape (including a columnar shape) and a random pattern depending on irregularity on the side surface is taken as a registration image, a reference sign or a reference mark for specifying a position of the registration image on the side surface may be hardly provided. In this case, the position of the registration image cannot be specified. For this reason, imaging for one circumference of the side surface of the elliptic cylindrical shape is forced to be performed.
Aspects of non-limiting embodiments of the present disclosure relate to a collation device, and a non-transitory computer readable medium storing a collation program that, in a case of imaging a side surface having an elliptic cylindrical shape with no reference sign or reference mark for specifying a position of a registration image and collating a captured image with the registration image, collate the captured image with the registration image to uniquely identify an object without imaging the side surface having the elliptic cylindrical shape for one circumference.
Aspects of certain non-limiting embodiments of the present disclosure overcome the above disadvantages and/or other disadvantages not described above. However, aspects of the non-limiting embodiments are not required to overcome the disadvantages described above, and aspects of the non-limiting embodiments of the present disclosure may not overcome any of the disadvantages described above.
According to an aspect of the present disclosure, there is provided a collation device including a memory that stores, as a registration image group, a random pattern depending on a microscopic pattern on a side surface of a target having an elliptic cylindrical shape including a columnar shape, and a processor, in which the registration image group is composed of a plurality of registration images respectively acquired at a plurality of positions at a predetermined interval over a whole circumference of the side surface, and the processor executes a program to collate a collation image obtained by imaging the side surface at an angle of view including at least one registration image among the plurality of registration images and the registration image group and to output a collation result.
Exemplary embodiment(s) of the present invention will be described in detail based on the following figures, wherein:
Hereinafter, an exemplary embodiment of the invention will be described based on the drawings.
In
The target 10 is a component that is subjected to various kinds of machining in the manufacturing process, and is a component having an elliptic cylindrical shape (including a columnar shape) in the exemplary embodiment. A specific example of the component having the elliptic cylindrical shape is a glass lens having a columnar shape, and a side surface of the glass lens is machined and processed in a ground glass shape. The glass lens having the columnar shape includes a convex glass lens or a concave glass lens that is represented by an aspheric lens. The registration image capturing machine 20 images a random pattern depending on microscopic irregularity on a side surface of the glass lens and registers a captured image as the registration image 16 in the collation device 24.
The target 10 succeeds to a pre-process 12 and is subjected to various kinds of processing in the post-process 14. In a transport process between the pre-process 12 and the post-process 14, sampling or lot confusion for the target 10 may occur. The same also applies to various processes of the post-process 14. At any timing of the post-process 14, a collation image capturing machine 22 images the random pattern depending on microscopic irregularity on the side surface of the glass lens as the target 10 and outputs a captured image as a collation image 23 to the collation device 24.
The collation device 24 collates the registration image 16 and the collation image 23 and determines whether or not both images coincide with each other. The collation device 24 outputs a collation result of the registration image 16 and the collation image 23 to a production management system 26. The collation device 24 calculates, for example, a degree of similarity of the registration image 16 and the collation image 23 and compares the calculated degree of similarity with a threshold value in size. In a case where the degree of similarity exceeds the threshold value, determination is made that the registration image 16 and the collation image 23 coincide with each other, and in a case where the degree of similarity does not exceed the threshold value, determination is made that the registration image 16 and the collation image 23 do not coincide with each other. Although the collation device 24 calculates the degree of similarity, for example, using a known collation algorithm, such as feature quantity matching depending on feature quantity detection or template matching using shading comparison of images, a calculation method of the degree of similarity is not limited thereto.
The production management system 26 acquires the collation result from the collation device 24 and reflects the collation result in a traceability state of the target 10. That is, the ID of the target 10 subjected to the pre-process 12 and the post-process 14 and an ID of a finished product in which the target 10 is incorporated are managed in association with each other.
In a case of collating the registration image 16 and the collation image 23 in the collation device 24, there is a need to make the position of the registration image 16 coincide with the position of the collation image 23 in the target 10. As a general method therefor, the following two methods are possible.
(1) The coordinates of the collation image 23 cut from the image obtained through imaging with the collation image capturing machine 22 are specified from a position of a pattern image in the image.
(2) The coordinates of the collation image 23 cut from the image obtained through imaging with the collation image capturing machine 22 are specified from absolute coordinates based on a designated position in the image.
Note that, since the method (1) supposes that a specific pattern to be a sign is present in the image, the method (1) cannot be used in a case where such a specific pattern to be a sign is not present. Furthermore, since the method (2) supposes that the designated position in the image can be specified, the method (2) cannot be used in a case where such a designated position is hardly specified.
In a case where the target 10 is a columnar glass lens with a side surface having a ground glass shape, since a specific pattern is not present on the side surface, and a position (an angle in a circumferential direction) of the side surface in an image is unclear, both methods (1) and (2) described above are hardly employed.
Of course, while a case where any mark, such as a minute flaw, is given to the glass lens having the columnar shape is also considered, this may affect the optical performance (refractive index and the like) of the glass lens.
In this way, in a case where there is no mark to be a sign in the target 10 or a mark to be a sign is hardly attached in the nature of the target 10, there is a need to image the whole circumference of the side surface of the glass lens for image collation.
Accordingly, in the exemplary embodiment, the registration images 16 are acquired at a plurality of positions at a predetermined interval over the whole circumference of the side surface of the glass lens as the target 10, and a plurality of registration images 16 are registered in the collation device 24. Then, in imaging the side surface of the glass lens with the collation image capturing machine 22, at least one registration image 16 among a plurality of registration images 16 is included within a range of an angle of view of the collation image capturing machine 22.
Then, in capturing a collation image with the collation image capturing machine 22, only one place of the side surface of the glass lens is imaged. In this case, in a case where the interval of a plurality of registration images 16 is appropriately set with respect to the angle of view of the collation image capturing machine 22, at least one registration image 16 among a plurality of registration images 16 is always included within the angle of view of the collation image capturing machine 22. For example, as shown in
Number n of registration images 16 =(Length of whole circumference of glass lens)/(Length of collation image 23−Length of registration image 16)
Here, the length of the collation image 23 or the length of the registration image 16 is defined as a size in a longitudinal direction (x direction) in
In a case where the number of registration images 16 is n, a range of searching with respect to the collation image is 1/n of the length of the whole circumference of the glass lens, and thus, calculation cost is reduced as much. Since collation of the collation image 23 obtained through imaging and the n registration images 16, that is, 1:n collation is performed, a fast computation algorithm can also be used directly. There is no need for detailed control or management, such as positioning of the glass lens in the circumferential direction with respect to the registration image capturing machine 20 and the collation image capturing machine 22 at the time of registration and collation of the registration images 16.
The collation device 24 includes a processor 24a, a ROM 24b, a RAM 24c, an input/output interface (I/F) 24d, a communication interface (I/F) 24e, and a registration image database (DB) 24f, which are connected to perform data transmission and reception via a bus 24g.
The processor 24a reads programs stored in the ROM 24b or other program memories and executes the programs using the RAM 24c as a working memory to execute various kinds of processing. The processing of the processor 24a is listed as follows.
The processor 24a receives a plurality of registration images 16 from the registration image capturing machine 20 via the communication I/F 24e and stores a plurality of registration images 16 in the registration image DB 24f. In this case, the ID of the glass lens as the target 10 and a plurality of registration images 16 are stored in the registration image DB 24f in association with each other.
The processor 24a receives the collation image 23 from the collation image capturing machine 22 via the communication I/F 24e and sequentially collates the collation image 23 with a plurality of registration images 16 stored in the registration image DB 24f.
The processor 24a stores a collation result of the registration image 16 and the collation image 23 in the registration image DB 24f and transmits the collation result to the production management system 26 through the communication I/F 24e.
The registration image DB 24f is configured with a storage device, such as a hard disk or a solid state disk (SSD), and stores a plurality of registration images. The registration image DB 24f functions as a memory that stores, as a registration image group, a random pattern depending on microscopic irregularity on the side surface of the target 10 having an elliptic cylindrical shape including a columnar shape.
Here, although the processor 24a sequentially compares a plurality of registration images 16 with one collation image 23, in a case where any one of the degrees of similarity of a plurality of registration images 16 and the collation image 23 exceeds the threshold value, the processor 24a determines that both images coincide with each other. On the other hand, in a case where all of the degrees of similarity of a plurality of registration images 16 and the collation image 23 do not exceed the threshold value, the processor 24a determines that both images do not coincide with each other. Specifically, in a case of the eight registration images 16a to 16h, in a case where the degree of similarity to any one of the registration images 16a to 16h exceeds the threshold value, the processor 24a determines coincidence, and in a case where all of the degrees of similarity to the registration images 16a to 16h do not exceed the threshold value, the processor 24a determines non-coincidence.
The collation device 24 may be configured with a personal computer (PC) or may be configured with a server computer. Although the registration image DB 24f is provided in the collation device 24 in
The processor 24a first acquires the registration images 16 of the glass lens as the target 10 (S101) . That is, the registration images 16 are acquired at a plurality of positions at predetermined intervals over the whole circumference of the side surface of the glass lens. For example, as shown in
Next, the processor 24a acquires the collation image 23 of the glass lens as the target 10 (S102). That is, the collation image 23 is acquired at any position on the side surface of the glass lens. Although the acquisition of the collation image 23 can be performed, for example, at any timing of the post-process 14 of the manufacturing process, the invention is not limited thereto, and the acquisition may be performed at a timing immediately after sampling or lot confusion of a product occurs. The processor 24a is connected to the collation image capturing machine 22 by the communication network, can transmit a control signal for instructing the collation image capturing machine 22 to capture the collation image, and can receive and acquire the collation image transmitted from the collation image capturing machine 22. Of course, the worker who manages the manufacturing process may operate the collation image capturing machine 22 to capture the collation image and to transmit the collation image to the collation device 24.
Next, the processor 24a collates a plurality of registration images 16 and the collation image 23 (S103). That is, the processor 24a calculates the degree of similarity of each of a plurality of registration images 16 and the collation image 23 and determines whether or not the calculated degree of similarity exceeds the threshold value.
Then, after performing image collation, the processor 24a outputs the collation result to a display device or the like via the input/output I/F 24d and outputs the collation result to the production management system 26 via the communication I/F 24e (S104). In a case of outputting the collation result to the display device, the worker who manages the manufacturing process can visually recognize the collation result to facilitate the management of the target 10.
Hereinafter, each kind of processing of
First, the glass lens as the target 10 is placed on the rotation stage 28 (S1011). In this case, a distance between the side surface of the glass lens and the registration image capturing machine 20 is adjusted, and a focal position of the registration image capturing machine 20 is focused on the side surface of the glass lens conforming to the side surface position of the glass lens.
Next, the glass lens as the target 10 is imaged while rotating by (length of collation image—length of registration image), and the registration images are cut (S1012). That is, the side surface of the glass lens is imaged with the registration image capturing machine 20, and the registration image having a predetermined size is cut. Thereafter, the rotation stage 28 is rotationally driven to rotate the glass lens by (length of collation image—length of registration image). The length of the registration image is a length that is set in advance and known. The length of the collation image is defined by the angle of view of the registration image capturing machine 20.
Then, the processing of S1012 is repeatedly executed until the side surface of the glass lens is rotated for one circumference, and a plurality n of registration images are acquired (S1013). Specifically, the following processing is repeated and the eight registration images 16a to 16h in total are acquired.
(1) capturing the registration image 16a
(2) rotating the glass lens by (length of collation image−length of registration image)
(3) capturing the registration image 16b
(4) rotating the glass lens by (length of collation image−length of registration image)
(5) capturing the registration image 16c
(6) rotating the glass lens by (length of collation image−length of registration image)
(7) capturing the registration image 16d
(8) rotating the glass lens by (length of collation image−length of registration image)
(9) capturing the registration image 16e
(10) rotating the glass lens by (length of collation image−length of registration image)
(11) capturing the registration image 16f
(12) rotating the glass lens by (length of collation image−length of registration image)
(13) capturing the registration image 16g
(14) rotating the glass lens by (length of collation image−length of registration image)
(15) capturing the registration image 16h
With this, the registration images 16a to 16h at regular intervals for (length of collation image−length of registration image) are acquired.
The acquired registration images 16a to 16h are stored in the registration image DB 24f in association with the ID of the glass lens. For example, in association with the ID of the glass lens, a data set (ID, registration image 16a, registration image 16b, registration image 16c, registration image 16d, registration image 16e, registration image 16f, registration image 16g, registration image 16h) is stored in the registration image DB 24f.
The registration image capturing machine 20 appropriately satisfies, for example, a condition that stable close-up imaging can be performed with high resolution and irradiation of the glass lens with light for objecting microscopic irregularity on the side surface of the glass lens can be stably performed.
Although the registration image DB 24f may be configured with a computer separated from the collation device 24 as described above, the ID of the glass lens as the target 10 and a plurality of acquired registration images 16a to 16h may be managed in a separate computer.
First, the glass lens as the target 10 is placed on the rotation stage 28 (S1021) . The rotation stage 28 in this case may be identical to or different from the rotation stage 28 in a case where the registration images 16 are acquired. Since the collation image 23 is captured from only one place, the rotation stage is not needed, and a non-rotation fixed stage may be used. In this case, a distance between the collation image capturing machine 22 and the rotation stage (or the fixed stage) is adjusted, and a focal position of the collation image capturing machine 22 is focused on the side surface of the glass lens conforming to the side surface of the glass lens. In more detail, the focal position of the collation image capturing machine 22 is focused on a cross point of a line segment that connects the collation image capturing machine 22 and the center of curvature of the glass lens having the columnar shape, and the side surface of the glass lens, that is, at a nearest position of the collation image capturing machine 22 and the glass lens.
Then, one place of the side surface of the glass lens as the target 10 is imaged with the collation image capturing machine 22, and one collation image 23 is cut (S1022) . The length of the collation image 23 is a length defined by the angle of view of the collation image capturing machine 22 and is greater than the length of the registration image 16. The width of the collation image 23 is also greater than the width of the registration image 16 (see
Since the interval of a plurality of registration images 16 is defined by (length of collation image−length of registration image), at least one registration image among a plurality of registration images 16a to 16h is included in one collation image 23 obtained through imaging at one place. For example, only the registration image 16e among a plurality of registration images 16a to 16h is included in one collation image 23. Alternatively, the registration images 16e and 16f among a plurality of registration images 16a to 16h are included in one collation image 23.
Similarly to the registration image capturing machine 20, the collation image capturing machine 22 appropriately satisfies, for example, a condition that stable close-up imaging can be performed with high resolution and irradiation of the glass lens with light for objecting microscopic irregularity on the side surface of the glass lens can be stably performed. Although the collation image capturing machine 22 is appropriately, for example, an identical model to the registration image capturing machine 20, the invention is not limited thereto as long as a substantially identical imaging condition to an imaging condition at the time of capturing the registration images is obtained.
First, a counter j for counting the glass lens group is initialized to j=1 (S201), and a registration image group of a j-th glass lens is referred to from the registration image DB 24f (S202). Next, a counter i for counting the registration images is initialized to i=1 (S203), and collation calculation of an i-th registration image and the collation image is performed and the degree of similarity is calculated (S204). That is, the registration images 16a to 16h are set as follows.
registration image 16a: first
registration image 16b: second
registration image 16c: third
registration image 16d: fourth
registration image 16e: fifth
registration image 16f: sixth
registration image 16g: seventh
registration image 16h: eighth
Then, the first registration image 16a is read from the registration image DB 24f to perform collation calculation with the collation image 23, and the degree of similarity of both images is calculated. In regard to the degree of similarity, a known algorithm can be used as described above.
In a case where the degree of similarity of the first registration image 16a and the collation image 23 is calculated, the calculated degree of similarity is compared with the threshold value, and determination is made whether or not the degree of similarity exceeds the threshold value (S205).
The image collation has an error rate due to fluctuation in an input of an image sensor of the registration image capturing machine 20 or the collation image capturing machine 22, a quantization error, or the like. The error rate is composed of two rates of an erroneous rejection rate that is a probability of determining as false despite true and an erroneous acceptance rate that is a probability of determining as true despite false. Both rates are in a relationship of trade-off, and in a case where one rate decreases, the other rate increases. Accordingly, the threshold value is set such that loss is minimized in an application target of collation determination.
As a result of size comparison of the degree of similarity with the threshold value, in a case where the degree of similarity exceeds the threshold value (in S205, YES), the processor 24a determines that the registration image 16a and the collation image 23 coincide with each other (S207). Then, the process progresses to processing of
On the other hand, in a case where the degree of similarity does not exceed the threshold value (in S205, NO), next, determination is made whether or not the counter i is equal to n (in a case of the registration images 16a to 16h, n=8) that is the number of registration images 16 (S206), and in a case where i does not yet reach n (in S206, NO), the counter i is incremented by 1 (S208), and the processing after S204 is repeated. That is, the second registration image 16b is read from the registration image DB 24f to perform collation calculation with the collation image 23, the degree of similarity of both images is calculated, determination is made whether or not the degree of similarity exceeds the threshold value, in a case where the degree of similarity exceeds the threshold value, determination is made that both images coincide with each other, and in a case where the degree of similarity does not exceed the threshold value, next, the degree of similarity of the third registration image 16c and the collation image is calculated.
The above-described processing is repeatedly executed until i=n (S206). Then, in a case where the degrees of similarity to all of the registration images 16 of the first registration image 16a to the eighth registration image 16h do not exceed the threshold value (in S206, YES), determination is made that both images do not coincide with each other (S209). This means that the glass lens specified by the registration image 16 and the glass lens acquired by the collation image 23 are different glass lenses. Then, the process progresses to processing of
In
In this way, a plurality of registration images 16 are prepared such that at least one registration image is always included in the collation image 23 obtained by imaging the side surface of the glass lens at one place, whereby there is no need to image the whole circumference of the side surface of the glass lens, and the registration image 16 and the collation image 23 can be collated to trace the identity of the glass lenses in the manufacturing process.
In the exemplary embodiment, although a plurality of registration images 16 are prepared such that at least one registration image is always included in the collation image 23 obtained by imaging the side surface of the glass lens at one place, a position where the registration image 16 is present in the collation image 23 is not determined, and the registration image 16 may be at the substantially center of the collation image 23 or the registration image 16 may be positioned in an end portion of the collation image 23. In a case of imaging the side surface of the glass lens as the target 10 with the collation image capturing machine 22, imaging is performed while the focusing position of the collation image capturing machine 22 conforms to the side surface of the glass lens, in more detail, the nearest distance of the side surface of the glass lens. For this reason, focusing is made at the substantially center of the angle of view of the collation image capturing machine 22; however, since the distance from the collation image capturing machine 22 increases depending on the curvature of the side surface of the glass lens in the end portion of the angle of view, focusing deviates from the focusing position, and the collation image is blurred.
Accordingly, the registration image 16 maybe positioned at the substantially center of the collation image 23, and the focused collation image 23 and the registration image 16 may be collated; however, in a case where the registration image 16 is positioned in the end portion of the collation image 23, and the non-focused and blurred collation image 23 and the registration image 16 are collated, correct collation calculation cannot be performed.
Accordingly, in a case where the registration image 16 is positioned at the substantially center of the collation image 23, the collation accuracy is secured; however, in a case where the registration image 16 is positioned at other positions, the collation accuracy is not secured, and as a result, a collatable range is restricted to the substantially center of the collation image.
Of course, since A is relatively decreased in a case where the curvature of the side surface of the glass lens as the target 10 is relatively small, the collation accuracy can be secured over the substantially entire region of the collation image; however, in a case where the curvature of the side surface is relatively large, a problem may occur.
In a case where a depth of field of the collation image capturing machine 22 is sufficiently deep, even though focusing deviates by the distance Δ, focusing can be kept within the focusing distance; however, in general, in imaging the random pattern of microscopic irregularity on the side surface of the glass lens, since imaging is performed at a close distance, the depth of field is relatively shallow, and in a case where focusing deviates by the distance Δ, focusing is out of the focusing position and the collation image 23 is often blurred.
Accordingly, in a first modification example, the focusing position of the collation image capturing machine 22 is shifted to a position entering inside the side surface of the glass lens from the side surface of the glass lens by a minute distance, not the side surface of the glass lens (more accurately, the position on the side surface of the glass lens at the nearest distance from the collation image capturing machine 22).
The collation accuracy is the maximum at angles +δ and −δ, and since the collation image 23 is blurred, the collation accuracy is degraded at the angle of 0 degrees. Ina case where the angle increases more than the angle +δ in the +direction, the collation accuracy is degraded, and in a case where the angle increases more than the angle −δ in the −direction, the collation accuracy is degraded similarly. Note that, since the collation accuracy is the maximum at the angles +δ and −δ, the degree of degradation of the collation accuracy at the angles +θ and −θ is small compared to the case of
In the exemplary embodiment, in a case where determination is made that all of the degrees of similarity to a plurality of registration images 16 do not exceed the threshold value, determination is made that the registration image 16 and the collation image 23 do not coincide with each other. In contrast, in the first modification example, the position of the registration image 16 in the collation image 23 is present, for example, in the end portion of the collation image 23, there is a possibility that the degree of similarity does not exceed the threshold value due to degradation of the collation accuracy caused by the blur of the collation image 23 in the end portion. In light of this, in a case where determination is made to be YES in S1034, the focusing position of the collation image capturing machine 22 is shifted from the nearest distance on the side surface of the glass lens inside the glass lens by a minute distance (S1038). A shift amount is any angle δ satisfying 0<δ<θ. After the focusing position is shifted, the processing after S1032 is repeated.
In the exemplary embodiment, the registration images 16 and the collation image 23 are collated with imaging of one place on the side surface of the glass lens. In contrast, in the first modification example, in a case where the registration image 16 and the collation image 23 are collated and do not coincide with each other with first imaging of one place on the side surface of the glass lens, the focal position is changed to perform second imaging of one place on the side surface of the glass lens, and the registration image 16 and the collation image 23 are collated.
There is no guarantee that the glass lens having the columnar shape as the target 10 is necessarily a perfect column, and variation in outer diameter may be present. Since deviation may occur in the positional relationship between the collation image capturing machine 22 and the glass lens, the collation image 23 maybe imaged in a focusing state, and image blur may be present.
Accordingly, in the second modification example, after the collation image 23 is acquired by imaging one place on the side surface of the glass lens, the glass lens is rotated by a predetermined angle, for example, (½)θ to capture the collation image 23.
In this case, as described in the first modification example, focusing may not be made and image blur may be present in both end portions of the collation image 23. Alternatively, focusing may not be made and image blur may be present in both end portions of the collation image 23 due to variation in outer diameter of the glass lens.
Accordingly, after the collation image 23 is acquired by imaging one place on the side surface of the glass lens, the glass lens is rotated by a predetermined angle, specifically, (½)θ.
(b) of
(a) of
(b) of
A difference from
In the exemplary embodiment, in a case where determination is made that all of the degrees of similarity to a plurality of registration images 16 do not exceed the threshold value, determination is made that the registration image 16 and the collation image 23 do not coincide with each other. In contrast, in the second modification example, the glass lens as the target 10 is rotated by a given angle, specifically, (½)θ (S1039). In a case where the glass lens is placed on the rotation stage 28, the glass lens can be rotated by rotationally driving the rotation stage 28. After the glass lens is rotated, the processing after S1032 is repeated.
In the exemplary embodiment, the registration images 16 and the collation image 23 are collated with imaging of one place on the side surface of the glass lens. In contrast, in the second modification example, in a case where the registration image 16 and the collation image 23 are collated and do not coincide with each other with first imaging of one place on the side surface of the glass lens, second imaging is performed at a different angle in the circumferential direction of the side surface of the glass lens, and the registration image 16 and the collation image 23 are collated.
In the second modification example, although the glass lens is rotated by (½)θ, a rotation angle is arbitrary, and the glass lens may be rotated by θ. While a flaw or a stain may be present on the side surface of the glass lens, and the collation accuracy may be degraded due to the flaw or the stain, even in this case, the glass lens may be rotated by a given angle to acquire the collation image 23. Specifically, in a case where the degrees of similarity to all registration images 16 do not exceed the threshold value in the first collation, the glass lens is rotated by θto perform second collation. Then, in a case where the degrees of similarity to all registration images 16 do not exceed the threshold value even in the second collation, the glass lens is rotated by θ again to perform third collation, and the like.
Of course, for example, an upper limit number of times of collation is set, and in a case where the degrees of similarity to all registration images 16 do not exceed the threshold value even though the upper limit number of times is reached, the determination processing ends, and finally, determination is appropriately made as non-coincidence.
Although the modification examples have been described above, the first modification example and the second modification example may be combined with each other. For example, like a case where the second modification example is executed after the first modification example is executed, any one of the first modification example and the second modification example is switched and executed depending on user's selection.
The exemplary embodiment and the first and second modification examples may be combined depending on the nature of the target 10. For example, the exemplary embodiment is executed to perform determination regarding a certain target 10, and any one of the first modification example and the second modification example is executed to perform collation a plurality of times from a viewpoint of improving the collation accuracy regarding another target 10.
In the exemplary embodiment and the modification examples, although a case where the collation image and the registration image group are collated has been described, a registration image having a length over the whole circumference of the side surface of the glass lens may be set as a registration image, and a plurality of collation images obtained by imaging a plurality of places on the side surface of the glass lens at a given angle of view and the registration image may be collated. Even with this, in a case of imaging the side surface having the elliptic cylindrical shape with no reference sign or reference mark for specifying the position of the registration image and collating the captured image with the registration images, collation with the registration images can be performed to perform unique identification without imaging the side surface having the elliptic cylindrical shape for one circumference.
In the embodiments above, the term “processor” refers to hardware in a broad sense. Examples of the processor include general processors (e.g., CPU: Central Processing Unit) and dedicated processors (e.g., GPU: Graphics Processing Unit, ASIC: Application Specific Integrated Circuit, FPGA: Field Programmable Gate Array, and programmable logic device) . In the embodiments above, the term “processor” is broad enough to encompass one processor or plural processors in collaboration which are located physically apart from each other but may work cooperatively. The order of operations of the processor is not limited to one described in the embodiments above, and may be changed.
The foregoing description of the exemplary embodiments of the present invention has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Obviously, many modifications and variations will be apparent to practitioners skilled in the art. The embodiments were chosen and described in order to best explain the principles of the invention and its practical applications, thereby enabling others skilled in the art to understand the invention for various embodiments and with the various modifications as are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the following claims and their equivalents.
Number | Date | Country | Kind |
---|---|---|---|
2021-151906 | Sep 2021 | JP | national |