Shoe processing assistance method, shoe processing assistance device, and shoe processing assistance program

Abstract

A shoe processing assistance device 14 comprises: a first acquisition unit 102 that acquires captured image data of a bonding surface of a first shoe component; a shape data generation unit 103 that generates first shape data indicating the shape of the bonding surface of the first shoe component on the basis of the captured image data; a first extraction unit 104 that extracts, on the basis of the first shape data, candidates of a processing area to be subjected to processing on the bonding surface for bonding to a surface to be bonded to of a second shoe component; a second acquisition unit 105 that acquires second shape data indicating the shape of the surface to be bonded to of the second shoe component; a second extraction unit 106 that extracts an area to be bonded to on the surface to be bonded to on the basis of the second shape data; and an identification unit 107 that identifies a processing area corresponding to each area to be bonded to among the candidates of the processing area on the basis of comparison of the candidates of the processing area and the area to be bonded to.

Description

Shoe processing support method, shoe processing support device, and shoe processing support program

[0001] The present disclosure relates to a shoe processing support method, a shoe processing support device, and a shoe processing support program.

[0002] Shoe manufacturing involves joining the midsole and outsole together, which involves processing such as buffing and applying adhesive to the joining surfaces between the midsole and outsole.

[0003] Meanwhile, automation of shoe manufacturing is progressing (see, for example, Patent Document 1). Patent Document 1 discloses a technique for attaching a shoe component to a base shoe component at a predetermined attachment position based on the shape of the shoe component.

[0004] US Patent Application Publication No. 2020 / 0229544

[0005] The technology described in Patent Document 1 cannot deal with manufacturing errors in the shoe components to be processed, and therefore there are cases where it is not possible to perform appropriate processing depending on the shoe component.

[0006] In view of the above, the present disclosure provides a technology that enables more appropriate processing to be performed depending on the shape of the joining surface of the shoe component to be processed.

[0007] A shoe processing support method according to one aspect of the present disclosure includes the steps of: acquiring image data of a joining surface of a first shoe component; generating first shape data indicating the shape of the joining surface of the first shoe component based on the image data; extracting, based on the first shape data, candidate processing areas to be processed for joining to a joining surface of a second shoe component that is to be joined to the first shoe component by the joining surface; acquiring second shape data indicating the shape of the joining surface of the second shoe component; extracting, based on the second shape data, the joining area that is to be joined to the first shoe component by the joining surface; and identifying, from the candidate processing areas, the processing areas that correspond to each of the joining areas based on a comparison between the candidate processing areas and the joining areas.

[0008] A shoe processing support device according to another aspect of the present disclosure includes a first acquisition unit that acquires imaging data of a joining surface of a first shoe component, a shape data generation unit that generates first shape data indicating a shape of the joining surface of the first shoe component based on the imaging data, a first extraction unit that extracts, based on the first shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is to be joined to the first shoe component by the joining surface, a second acquisition unit that acquires second shape data indicating the shape of the joining surface of the second shoe component, a second extraction unit that extracts, based on the second shape data, the joining area that is to be joined to the first shoe component by the joining surface, and an identification unit that identifies the processing area among the candidate processing areas that corresponds to each of the joining areas based on a comparison between the candidate processing areas and the joining areas.

[0009] A shoe processing support program according to yet another aspect of the present disclosure causes a computer to execute the following steps: acquiring imaging data of a joining surface of a first shoe component; generating first shape data based on the imaging data, the first shape data indicating the shape of the joining surface of the first shoe component; extracting, based on the first shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is to be joined to the first shoe component by the joining surface; acquiring second shape data indicating the shape of the joining surface of the second shoe component; extracting, based on the second shape data, the joining area that is to be joined to the first shoe component by the joining surface; and identifying, from the candidate processing areas, the processing areas that correspond to each of the joining areas based on a comparison between the candidate processing areas and the joining areas.

[0010] Any combination of the above components, and conversion of the present disclosure into a method, device, system, computer program, data structure, recording medium, etc., are also valid aspects of the present disclosure.

[0011] According to the present disclosure, a technology can be provided that enables more appropriate processing to be performed depending on the shape of the joining surface of the shoe component to be processed.

[0012] 1 is a schematic diagram showing the overall configuration of a shoe processing system according to an embodiment; FIG. 2 is a functional block diagram of a shoe processing support device of a first embodiment; FIG. 3 is a flowchart showing processing for generating processing data for the shoe processing support device of the first embodiment; FIG. 4 is a diagram showing an example of midsole shape data; FIG. 5 is a diagram showing examples of processing area candidates extracted from the midsole shape data of FIG. 4; FIG. 6 is a diagram showing an example of outsole shape data; FIG. 7 is a diagram showing examples of bonded areas extracted from the outsole shape data of FIG. 6; FIG. 8 is a diagram showing examples of center of gravity positions of an outsole and a midsole and position vectors of each center of gravity position; FIG. 9 is a diagram showing unit vectors of the second moment of area of ​​each bonded area and each candidate processing area; FIG. 10 is a diagram showing an example of a processing area on a joint surface of a midsole; FIG. 11 is a diagram showing an example of a processing trajectory in a processing area; FIG. 12 is a diagram showing examples of second moment of area of ​​each bonded area and each candidate processing area calculated for outsoles and midsole of different sizes; 20. is a flowchart showing processing for generating processing data for the shoe processing support device of the third embodiment. FIG. 21 is a diagram for explaining the flow of the shape correction processing of the third embodiment. FIG. 22 is a flowchart showing processing for generating processing data for the shoe processing support device of the fourth embodiment. FIG. 23 is a diagram showing the accuracy rate of matching when the processing area is specified using each specification method. FIG. 24 is a diagram illustrating the midsole shape data and outsole shape data used in FIG. 20. FIG. 25 is a functional block diagram of the shoe processing support device of the fifth embodiment. FIG. 26 is a flowchart showing processing for generating processing data for the shoe processing support device of the fifth embodiment. FIG. 27 is a diagram illustrating the likelihood that an area that is a candidate for the processing area is a processing area. FIG. 28 is a diagram for explaining another example of the method of specifying the center of gravity position by the specification unit and the method of narrowing down the candidates for the processing area.

[0013] The present disclosure will be described below based on preferred embodiments with reference to the accompanying drawings. In the embodiments and modifications, the same or equivalent components are designated by the same reference numerals, and redundant descriptions will be omitted as appropriate.

[0014] In the following, an example will be described in which the first shoe component is a midsole and the second shoe component joined to the first shoe component is an outsole, but the first shoe component and the second shoe component may be shoe components joined to each other, and are not limited to soles. For example, the first shoe component may be the main upper material of the upper and the second shoe component may be the reinforcing material of the upper, or the first shoe component may be the sole and the second shoe component may be a truss.

[0015] First Embodiment Fig. 1 is a schematic diagram showing the overall configuration of a shoe processing system 1 according to a first embodiment. The shoe processing system 1 is a system for processing shoes including an outsole 6 and a midsole 8. The shoe processing system 1 includes a camera 10, a shoe processing support device 14, a buffing device 16, and an application device 18. Here, "processing" refers to processes carried out prior to joining the outsole 6 and midsole 8, such as buffing and application processes described below.

[0016] The outsole 6 is placed on a table (not shown) by a worker.

[0017] The midsole 8 is attached to a holding stand (not shown) by an operator. The midsole 8 is held upside down with the joining surface (bottom surface) 8a, to which the outsole 6 is to be joined, facing upward. The camera 10 photographs the joining surface 8a of the midsole 8 held on the holding stand.

[0018] In step (A), the camera 10 captures an image of the joint surface 8a of the midsole 8. In the first embodiment, the camera 10 captures an image of the joint surface 8a of the midsole 8 from multiple viewpoints. This is not limiting, and the camera 10 may capture an image of the joint surface 8a of the midsole 8 from a single viewpoint. Note that a scanner such as a three-dimensional scanner may be used instead of the camera 10. Hereinafter, the image of the joint surface 8a of the midsole 8 captured by the camera 10 will be referred to as a midsole image.

[0019] In step (B), the buffing device 16 buffs the midsole 8 based on a control signal from the shoe processing support device 14. "Buffing" refers to a process of polishing and roughening the surface. The buffing device 16 includes a buffing unit 20 and a moving mechanism 22. The buffing unit 20 includes a polishing member such as a grindstone or a brush, and performs buffing on the midsole 8. The moving mechanism 22 moves the buffing unit 20. The moving mechanism 22 may be an articulated (multi-axis) robot arm. The moving mechanism 22 moves the buffing unit 20 according to a trajectory indicated by trajectory data described below. This causes the joining surface 8a, which is the area surface inside the boundary, to be buffed (i.e., the joining surface 8a becomes rough), making it easier for the adhesive to adhere to the joining surface 8a.

[0020] In step (C), the applicator 18 applies adhesive to the midsole 8 based on a control signal from the shoe processing support device 14. The applicator 18 includes an applicator 24 and a moving mechanism 26. The applicator 24 discharges adhesive from a discharge port provided at its tip. The moving mechanism 26 is a mechanism for moving the applicator 24. The moving mechanism 26 may be an articulated (multi-axis) robot arm. The moving mechanism 26 moves the applicator 24 according to a trajectory indicated by trajectory data described below. In this way, the adhesive is applied to the joining surface 8a that has been buffed.

[0021] In step (D), an operator manually bonds the outsole 6 to the midsole 8 having the adhesive applied to its joining surface 8a.

[0022] As a modified example, the shoe processing system 1 may be configured not to include the buffing device 16. For example, if the midsole 8 is made of a material that adhesives easily adhere to, there is no need to buff the midsole 8, and therefore the shoe processing system 1 does not need to include the buffing device 16.

[0023] In addition, in this example, the shoe processing support device 14 directly controls the buffing device 16 and the application device 18, but a PLC (programmable logic controller) may be provided separately from the shoe processing support device 14 and these devices may be controlled via the PLC.

[0024] The above is the overall configuration of the shoe processing system 1. Next, the shoe processing support device 14 will be described in detail.

[0025] 2 is a functional block diagram of the shoe processing support device 14 of the first embodiment. Each block shown here can be realized in hardware by elements and mechanical devices, such as a computer's CPU (Central Processing Unit), and in software by a computer program, but the functional blocks shown here are those realized by the cooperation of these elements. Therefore, those skilled in the art who have read this specification will understand that these functional blocks can be realized in various ways by combining hardware and software.

[0026] The shoe processing support device 14 includes a memory unit 101, a first acquisition unit 102, a shape data generation unit 103, a first extraction unit 104, a second acquisition unit 105, a second extraction unit 106, an identification unit 107, a processing data generation unit 108, a buffing processing control unit 109, and an application control unit 110.

[0027] The storage unit 101 stores a program for executing the processing of the shoe processing support device 14. The storage unit 101 also stores in advance outsole shape data that indicates the shape of the joining surface 6a of the outsole 6 to which the midsole 8 is to be joined. The storage unit 101 may store in advance outsole shape data generated by 3D-CAD or the like, or may store in advance outsole shape data generated by actually photographing the joining surface 6a of the outsole 6 with a camera.

[0028] The first acquisition unit 102 acquires imaging data of the joint surface 8a of the midsole 8. The first acquisition unit 102 of the first embodiment acquires a midsole image from the camera 10 as imaging data.

[0029] The shape data generation unit 103 generates midsole shape data indicating the shape of the joint surface 8a of the midsole 8 based on the midsole image. Here, the shape data is assumed to be point cloud data, but is not limited to this and may be, for example, mesh data. The configuration of the shape data generation unit 103 is not particularly limited and may be configured using known or future available technology. The midsole shape data of the first embodiment is an example of first shape data.

[0030] The first extraction unit 104 extracts candidate processing areas 60 (see FIG. 5 , described later) on the joint surface 8 a of the midsole 8, where processing is performed for joining to the outsole 6, based on the midsole shape data. The first extraction unit 104 of the first embodiment extracts candidate processing areas on the joint surface 8 a based on the shape of the outline of the joint surface 8 a of the midsole 8 and the shape of the rib 50 (see FIG. 4, described later) formed on the joint surface 8 a of the midsole 8. For example, the first extraction unit 104 extracts, as candidate processing areas, areas surrounded by at least one of the outline of the joint surface 8 a of the midsole 8 and the rib 50, the area of ​​which is equal to or greater than a predetermined area. The first extraction unit 104 may extract the area surrounded by at least one of the outline of the joint surface 8 a of the midsole 8 and the rib using known or future available technology.

[0031] The second acquisition unit 105 acquires outsole shape data. The second acquisition unit 105 in the first embodiment acquires the outsole shape data by reading outsole shape data from the storage unit 101. In the first embodiment, the outsole shape data is acquired for the outsole 6 specified by a user input to a user input device (not shown).

[0032] The second extraction unit 106 extracts a bonded region 40 (see FIG. 7 described later) of the outsole 6 that is to be bonded to the midsole 8, based on the outsole shape data. For example, the second extraction unit 106 extracts an area surrounded by the outline of an outsole member 30 (see FIG. 6 described later) of the outsole 6 as the bonded region 40. The second extraction unit 106 may extract the area surrounded by the outline of the outsole member 30 of the outsole 6 using known or future available technology.

[0033] The identifying unit 107 identifies a processing area 65 (see FIG. 10 described later) among the processing area candidates 60 that corresponds to each of the to-be-welded areas 40 based on a comparison between the processing area candidates 60 and the to-be-welded areas 40. A method for identifying the processing area 65 will be described later.

[0034] The processing data generation unit 108 generates various processing data for processing the processing area of ​​the midsole 8 based on the midsole shape data. For example, for each identified processing area 65, the processing data generation unit 108 generates processing area data that indicates a processing area 70 (see FIG. 10 described below) having an outer periphery that is offset inward by a predetermined distance from the outer periphery of the processing area 65 identified in the midsole shape data. For example, the processing data generation unit 108 generates trajectory data that indicates a trajectory when processing the processing area based on the processing area data. The processing data generation unit 108 may generate the trajectory data using known or future available technology.

[0035] The buffing control unit 109 controls the buffing device 16 based on the trajectory data generated by the processing data generation unit 108, and buffs the joining surface 8 a in the processing area 70. More specifically, the buffing control unit 109 controls the moving mechanism 22 so that the buffing device 20 moves according to the trajectory in the trajectory data.

[0036] The application control unit 110 controls the buffing device 16 based on the trajectory data generated by the processing data generation unit 108, and applies adhesive to the joining surface 8a in the processing area 70. More specifically, the buffing control unit 109 controls the movement mechanism 26 so that the application unit 24 moves according to the trajectory in the trajectory data.

[0037] FIG. 3 is a flowchart showing the process S100 for generating processing data for the shoe processing support device 14 of the first embodiment.

[0038] In step S101, the first acquisition unit 102 acquires a midsole image.

[0039] In step S102, the shape data generation unit 103 generates midsole shape data based on the midsole image. FIG. 4 shows an example of midsole shape data. In the example of FIG. 4, a central rib 50a is formed on the joint surface 8a of the midsole 8. The central rib 50a is arranged to extend in the foot length direction (y direction in FIG. 8 ) at the center of the joint surface 8a in the foot width direction (x direction in FIG. 8 , which will be described later). The central rib 50a is formed with foot width direction ribs 50b to 50f extending from the central rib 50a toward the outer edge of the joint surface 8a in the foot width direction, cross ribs 50g to 50j extending from the central rib 50a in the foot width direction to cross the outer edge of the joint surface 8a, and a plurality of inner ribs 50k formed within the area surrounded by the central rib 50a. Furthermore, outer foot ribs 50l and 50m are formed on the joint surface 8a of the midsole 8, extending in the foot width direction from the outer contour of the joint surface 8a on the outer foot side toward the central rib 50a. In this specification, the ribs 50a to 50m will be collectively referred to as rib 50 unless otherwise distinguished.

[0040] Here, the rib 50 will be described. A groove (recess) is formed in the midsole 8, and the protrusion that follows the outline of the groove is the rib 50. The rib 50 is provided, for example, to define a boundary for positioning the outsole 6 and the midsole 8. Furthermore, for example, the rib 50 has the role of suppressing misalignment of the outsole 6 when the outsole 6 is joined to the midsole 8.

[0041] In step S103, the first extraction unit 104 extracts candidate processing areas 60 on the joint surface 8a based on the midsole shape data. FIG. 5 illustrates candidate processing areas 60 extracted by the first extraction unit 104 from the midsole shape data of FIG. 4. FIG. 5 shows areas 60a-60e, 61, and 62. Area 60a is an area surrounded by the outer contour of the joint surface 8a, the central rib 50a, the foot width direction ribs 50b-50f, the cross ribs 50g and 50h, and the outer foot ribs 50l and 50m. Area 60b is an area surrounded by the outer contour of the joint surface 8a, the central rib 50a, and the cross ribs 50i and 50j. Area 60c is an area surrounded by the outer contour of the joint surface 8a, the central rib 50a, and the cross ribs 50h and 50i. Region 60d is an area surrounded by the outer contour of the joint surface 8a, the central rib 50a, and the transverse ribs 50g and 50j. Region 60e is an area surrounded by the central rib 50a, the foot width direction ribs 50b to 50f, and the inner rib 50k located closest to the toes. Region 61 is an area formed inside the transverse rib 50g. Region 62 is an area surrounded by the central rib 50a and multiple inner ribs 50k.

[0042] 5, the areas 60a to 60e are assumed to be areas whose areas are equal to or larger than a predetermined area. In this case, the first extraction unit 104 extracts the areas 60a to 60e as candidates for the first to fifth machining areas (hereinafter referred to as the first to fifth machining area candidates 60a to 60e). In this specification, when the first to fifth machining area candidates 60a to 60e are not particularly distinguished from each other, they are collectively referred to as the machining area candidate 60.

[0043] In step S104, the second acquisition unit 105 acquires outsole shape data. FIG. 6 shows an example of outsole shape data. In the example of FIG. 6, the outsole 6 is composed of first to fourth outsole members 30a to 30d. The first outsole member 30a is disposed on the toe side of the outsole 6. The second outsole member 30b is disposed on the heel side of the outsole 6. The third outsole member 30c is disposed on the outer side of the foot between the first outsole member 30a and the second outsole member 30b. The fourth outsole member 30d is disposed on the inner side of the foot between the first outsole member 30a and the second outsole member 30b. In this specification, the first to fourth outsole members 30a to 30d will be collectively referred to as outsole members 30 unless a distinction is made between them.

[0044] In step S105, the second extraction unit 106 extracts a bonded region 40 on the bonded surface 6a of the outsole 6 based on the outsole shape data. FIG. 7 illustrates the bonded region 40 extracted by the second extraction unit 106 from the outsole shape data of FIG. 6. As shown in FIG. 7, the second extraction unit 106 extracts first to fourth bonded regions 40a to 40d on the bonded surface 6a of the outsole 6 to be bonded to the bonded surface 8a of the midsole 8 based on the outsole shape data of FIG. 6. The first bonded region 40a is located on the toe side of the outsole 6 to correspond to the first outsole member 30a. The second bonded region 40b is located on the heel side of the outsole 6 to correspond to the second outsole member 30b. The third bonded region 40c is located on the outer side of the foot between the first outsole member 30a and the second outsole member 30b to correspond to the third outsole member 30c. The fourth bonded region 40d is disposed on the medial side of the foot between the first outsole member 30a and the second outsole member 30b so as to correspond to the fourth outsole member 30d. In this specification, the first to fourth bonded regions 40a to 40d will be collectively referred to as bonded region 40 unless otherwise specified.

[0045] Typically, the outsole 6 is discretized into multiple outsole members 30 as shown in FIG. 6 , and the joining surface 8a of the midsole 8 has a complex shape with multiple ribs 50 as shown in FIG. 4 . Therefore, the candidate processing areas 60 and the bonded areas 40 extracted by the first and second extraction units 106 are also discretized into multiple parts (see FIGS. 5 and 7 ). The number, shape, and arrangement of the candidate processing areas 60 and the bonded areas 40 vary significantly depending on the type of midsole 8 and outsole 6. Therefore, it is necessary to identify the processing areas corresponding to each bonded area 40 by appropriately matching the extracted candidate processing areas 60 with the extracted bonded areas 40. The following steps S106 to S110 illustrate an example of a method for identifying the processing areas corresponding to each bonded area 40.

[0046] In step S106, the identification unit 107 identifies the center of gravity positions on the xy plane of the edge curves of the bonded surface 6a of the outsole 6, the first to fourth bonded regions 40a to 40d, the bonded surface 8a of the midsole 8, and the first to fifth candidate processing regions 60a to 60e. FIG. 8 illustrates the center of gravity positions of the outsole 6 and the midsole 8 and the position vectors of each center of gravity position described below. In FIG. 8, the x direction is the foot width direction, and the y direction is the foot length direction. In the example of FIG. 8, the identification unit 107 identifies the center of gravity position of the entire bonded surface 6a of the outsole 6 as GOi, the center of gravity positions of the first to fourth bonded regions 40a to 40d as GOa to GOd, the center of gravity position of the entire bonded surface 8a of the midsole 8 as GMi, and the center of gravity positions of the first to fifth candidate processing regions 60a to 60e as GMa to GMe.

[0047] In step S107, the identification unit 107 calculates the position vectors of the center of gravity positions GOa to GOd when the center of gravity position GOi is used as a reference, and the position vectors of the center of gravity positions GMa to GMe when the center of gravity position GMi is used as a reference. In the example of Fig. 8, the identification unit 107 calculates the position vectors from the center of gravity position GOi to each of the center of gravity positions GOa to GOd as position vectors VOa to VOd, respectively, and calculates the position vectors from the center of gravity position GMi to each of the center of gravity positions GMa to GMe as position vectors VMa to VMe.

[0048] In step S108, the identifying unit 107 narrows down the candidates for the first to fifth machining regions 60a to 60e corresponding to the first to fourth welded regions 40a to 40d, respectively, based on the calculated position vectors VOa to VOd and VMa to VMe. For example, to narrow down the candidates for the machining region corresponding to the first welded region 40a, the identifying unit 107 calculates the angles formed by each of the position vectors VMa to VMe with respect to the position vector VOa, and extracts the candidates for the machining region 60 having position vectors whose calculated angles are equal to or less than a predetermined angle threshold. In this embodiment, the identifying unit 107 narrows down the candidates for the machining region 60 corresponding to the first welded region 40a to the candidates for the first and fifth machining regions 60a and 60e. Similarly, the identifying unit 107 calculates the angle between each of the position vectors VMa to VMe and each of the position vectors VOb to VOd for the second to fourth welded regions 40b to 40d, and extracts candidates for processing regions having position vectors whose calculated angles are equal to or less than a predetermined angle threshold, thereby narrowing down the candidate processing regions 60 corresponding to each of the second to fourth welded regions 40b to 40d. In this embodiment, the identifying unit 107 narrows down the candidates for processing regions corresponding to the second to fourth welded regions 40b to 40d to the candidates for the second to fourth processing regions 60b to 60d.

[0049] In step S109, the identification unit 107 calculates the second moments of area Ix and Iy in the x and y directions of the edge curves of the first to fourth welding regions 40a to 40d and the first to fifth processing region candidates 60a to 60e using the following equations (1) and (2). dA in equations (1) and (2) represents an infinitesimal cross-sectional area.

[0050] In step S109, the specifying unit 107 calculates a unit vector Iu having elements Ixu and Iyu from a vector I having elements Ix and Iy, for each of the first to fourth welded regions 40a to 40d and the first to fifth candidate processing regions 60a to 60e. For example, the specifying unit 107 calculates Ixu based on the elements Ix and Iy. ２ +Iyu ２ A unit vector Iu having Ixu and Iyu as elements is calculated so that Iu=1.

[0051] In step S110, the identifying unit 107 identifies the processing area 65 corresponding to the welded area 40 based on the similarity of the shapes of each welded area 40 and each candidate processing area 60. For example, the identifying unit 107 identifies the candidate processing area 60 having the smallest deviation between the second moments of area in the x direction and the y direction of the candidate processing area 60 and the second moments of area in the x direction and the y direction of the welded area 40 as the processing area corresponding to the welded area 40. Here, the unit vector of the vector I having the second moments of area Ix and Iy for the ith candidate processing area 60 as elements is defined as Ixu i,pro and Iyu i,pro A unit vector Iu with elements i,pro The unit vector of vector I, whose elements are the second moments of area Ix and Iy for the j-th welded region 40, is expressed as Ixu j,bond and Iyu j,bond A unit vector Iu with elements j,bond For example, the specifying unit 107 may select a unit vector Iu of a vector I having the second moments of area Ix and Iy as elements from among the candidates 60 for the processing region narrowed down for each of the welded regions 40. i,pro Element Ixu i,pro and Iyu i,pro and a unit vector Iu of a vector I having the second moments of area Ix and Iy of each welded region 40 as elements. j,bond Element Ixu j,bond and Iyu j,bond Based on this, the Euclidean distance between the i-th candidate processing region 60 and the j-th welded region 40 is calculated. The Euclidean distance here is an example of "the deviation between the second moments of area in the x and y directions of the candidate processing region 60 and the second moments of area in the x and y directions of the welded region 40." The Euclidean distance d i,j is calculated using the following formula (3).

[0052] The identification unit 107 identifies the candidate processing area 60 with the smallest calculated Euclidean distance from among the narrowed-down candidates for processing area 60 for each bonded area 40 as the processing area corresponding to that bonded area 40. By making a comparison using unit vectors, it becomes possible to appropriately identify the processing area even if the size of the captured midsole 8 differs from the size of the outsole 6 stored in advance as outsole shape data.

[0053] FIG. 9 illustrates elements Ixu and Iyu of a unit vector Iu of a vector I, whose elements are the second moments of area of ​​each welded region 40 and each candidate for a processing region 60. The candidate processing regions 60 corresponding to the second to fourth welded regions 40b to 40d are narrowed down to one each (the second to fourth candidate processing regions 60b to 60d) in step S108. Therefore, the identifying unit 107 identifies the second to fourth candidate processing regions 60b to 60d as processing regions 65b to 65d (see FIG. 10 described below) corresponding to the welded regions 40b to 40d. Furthermore, the candidate processing region 60 corresponding to the first welded region 40a is narrowed down to two candidates, the first and fifth candidate processing regions 60a and 60e, in step S108. Here, the Euclidean distance d between the candidate for the first processing region 60a and the first welded region 40a is 1,1 is 0.0264, and the Euclidean distance d 5,1 Therefore, the identifying unit 107 identifies the first processing region candidate 60a, which has the smallest Euclidean distance, from among the first and fifth processing region candidates 60a and 60e narrowed down for the first welded region 40a in step S108, as the processing region 65a (see FIG. 10 described later) corresponding to the first welded region 40a.

[0054] In step S111, the processing data generation unit 108 generates processing area data. FIG. 10 illustrates a processing area 70 on the joint surface 8a of the midsole 8. FIG. 10 shows processing areas 70a-70d, each having an outer periphery offset a predetermined distance inward from the outer periphery of the processing areas 65a-65d. The "predetermined distance" here is, for example, a distance corresponding to the width of the abrasive in the buffing unit 20. For example, the "predetermined distance" is a distance equal to or greater than the width of the abrasive in the buffing unit 20. In this specification, when there is no particular need to distinguish between the processing areas 65a-65d and the processing areas 70a-70d, they will be collectively referred to as the processing areas 65 and the processing areas 70, respectively.

[0055] In step S112, the processing data generation unit 108 generates trajectory data. Fig. 11 shows an example of a processing trajectory in the processing region 70 indicated by the trajectory data. Fig. 11 shows multiple trajectories 80 along the outlines of the processing regions 70a to 70d, so that processing is performed at multiple positions on each processing region 70.

[0056] After step S112, the process S100 ends.

[0057] However, in the technology described in Patent Document 1, the shape of the processing area of ​​the shoe component to be processed depends on the shape of the joining area, making it difficult to deal with manufacturing errors in the shape of the shoe component to be processed, etc. As a result, there are cases where it is not possible to perform appropriate processing depending on the shoe component.

[0058] In contrast, the shoe processing support device 14 of the first embodiment includes: a first acquisition unit 102 that acquires imaging data of the joining surface 8 a of the midsole 8; a shape data generation unit 103 that generates midsole shape data indicating the shape of the joining surface 8 a of the midsole 8 based on the imaging data; a first extraction unit 104 that extracts, based on the midsole shape data, candidate processing areas 60 to be processed for joining to a joining surface 6 a of the outsole 6 that is joined to the midsole 8 by the joining surface 8 a of the midsole 8; a second acquisition unit 105 that acquires outsole shape data indicating the shape of the joining surface 6 a of the outsole 6; a second extraction unit 106 that extracts, based on the outsole shape data, the joining areas 40 that are joined to the midsole 8 by the joining surface 6 a of the outsole 6; and an identification unit 107 that identifies, based on a comparison between the candidate processing areas 60 and the joining areas 40, the processing areas 65 that correspond to each of the joining areas 40 among the candidate processing areas 60. According to this configuration, the processing area 65 on the joint surface 8a of the midsole 8 is identified from the image data of the joint surface 8a of the midsole 8. Therefore, it becomes possible to perform an appropriate processing treatment on the joint surface 8a of the midsole 8 depending on the manufacturing error of the midsole 8, etc.

[0059] Here, since the polishing member has a width, if the buffing device 16 is moved so that the center of the polishing member passes through the periphery of the processing area 65, buffing will occur beyond the periphery of the processing area 65. Therefore, as shown in the first embodiment, the processing data generation unit 108 may generate processing area data for each identified processing area 65, which indicates a processing area 70 having an outer periphery that is offset inward by a predetermined distance from the periphery of the processing area 65. Furthermore, the processing data generation unit 108 may generate trajectory data indicating a trajectory when processing the processing area 70. This prevents processing from occurring beyond the periphery of the processing area 65.

[0060] As shown in the first embodiment, the first extraction unit 104 may extract the candidate processing area 60 based on the shape of the outer contour of the joint surface 8a of the midsole 8 and the shape of the rib 50 formed on the joint surface 8a of the midsole 8. With this configuration, it is possible to appropriately extract the candidate processing area 60.

[0061] As shown in the first embodiment, the second extraction unit 106 may extract the bonded region 40 based on the outsole shape data stored in advance in the storage unit 101. According to this configuration, there is no need to actually photograph the bonded surface 6 a of the outsole 6 in the process for generating the processing data of the shoe processing support device 14 as described above, which can contribute to improving work efficiency and reducing work costs.

[0062] As a comparative example, FIG. 12 illustrates the second moments of area Ix and Iy of each bonded region 40 and each candidate processing region 60 calculated for an outsole 6 and a midsole 8 of different sizes. As shown in FIG. 12 , the size Sout of the outsole 6 is 27.0 centimeters. The size Smid of the midsole 8 is 29.0 centimeters. In this case, if an attempt is made to identify the processing region 65 corresponding to each bonded region 40 based on the Euclidean distance between vectors I each having the second moments of area Ix and Iy of each bonded region 40 and each candidate processing region 60 as elements, the Euclidean distance between vectors I each having the second moments of area Ix and Iy of each corresponding region of the outsole 6 and midsole 8 as elements will differ significantly due to the size difference. As a result, it may be difficult to identify the processing region 60. Therefore, as shown in the first embodiment, the identification unit 107 may correct the size of the to-be-bonded region 40 in the outsole shape data based on the size of the bonding surface 8 a of the midsole 8 and the size of the to-be-bonded surface 6 a of the outsole 6 stored in advance in the storage unit 101. For example, the identification unit 107 may correct the size of the to-be-bonded region 40 in the outsole shape data by calculating a unit vector Iu of a vector I whose elements are the second moments of area Ix and Iy in the x and y directions of the edge curves of the to-be-bonded region 40 and the candidate processing region 60. With this configuration, the processing region 65 can be appropriately identified even if the sizes of the outsole 6 and the midsole 8 are different. Therefore, it is only necessary to register one piece of outsole shape data for each product type, which makes it possible to suppress an increase in the amount of data.

[0063] Second Embodiment A second embodiment of the present disclosure will be described below. In the drawings and description of the second embodiment, components and members that are the same as or equivalent to those in the first embodiment will be denoted by the same reference numerals. Explanations that overlap with the first embodiment will be omitted as appropriate, and the description will focus on the configurations that differ from the first embodiment.

[0064] 13 is a functional block diagram of the shoe processing support device 14 of the second embodiment. The shoe processing support device 14 of the second embodiment further includes an acceptance / rejection determination unit 111. The acceptance / rejection determination unit 111 determines whether the shape of the joining surface 8a of the midsole 8 is acceptable or not based on the shape of the identified processing area 65 and the shape of the joining area 40 corresponding to the processing area 65.

[0065] Fig. 14 is a flowchart showing the process S200 for generating processing data for the shoe processing support device 14 of the second embodiment. In the process S200 of Fig. 14, steps S201 to S210 and S213 to S214 are basically the same as steps S101 to S112 of Fig. 3 except where specifically mentioned, and therefore a description thereof will be omitted.

[0066] In step S211, the pass / fail determination unit 111 determines whether the shape of the joining surface 8a of the midsole 8 passes or fails based on the shape of each identified processing region 65 and the shape of the bonded region 40 corresponding to that processing region 65. For example, the pass / fail determination unit 111 determines whether the deviation between the second moment of area in each of the x and y directions of the processing region 65 and the second moment of area in each of the x and y directions of the bonded region 40 corresponding to that processing region 65 is equal to or less than a threshold. More specifically, for example, the pass / fail determination unit 111 determines whether the Euclidean distance calculated for each processing region 65 and the bonded region 40 corresponding to that processing region is equal to or less than a threshold. If the Euclidean distance is equal to or less than the threshold (Yes in step S211), the shape of the joining surface 8a of the midsole 8 is determined to pass, and the process S200 proceeds to step S213. If the Euclidean distance is not equal to or less than the threshold value (No in step S211), the shape of the joint surface 8a of the midsole 8 is determined to be unacceptable, and the process S200 proceeds to step S212.

[0067] In step S212, the pass / fail determination unit 111 reports a failure determination regarding the shape of the joint surface 8a of the midsole 8. For example, the pass / fail determination unit 111 reports the failure using a display device or audio output device (not shown) of the shoe processing support device 14. After step S212, the process S200 ends.

[0068] As shown in the second embodiment, the pass / fail determination unit 111 may determine the pass / fail of the shape of the joining surface 8a of the midsole 8 based on the shape of the identified processing area 65 and the shape of the joining area 40 corresponding to the processing area 65. This configuration makes it possible to accurately select midsoles 8 having joining surfaces 8a with appropriate shapes.

[0069] Third Embodiment A third embodiment of the present disclosure will now be described. In the drawings and description of the third embodiment, components and members that are the same as or equivalent to those in the first embodiment will be denoted by the same reference numerals. Explanations that overlap with the first embodiment will be omitted as appropriate, and the description will focus on configurations that differ from the first embodiment.

[0070] FIG. 15 shows an example of outsole shape data and midsole shape data. In the example of FIG. 15, processing regions 65f-65m correspond to bonding regions 40f-40m, respectively. In the midsole shape data shown in FIG. 15, distorted portions 80f-80h may be formed in the contours of processing regions 65f-65h due to data noise or the like. In the third embodiment, a method for correcting these distorted portions 80f-80h to an appropriate shape will be described.

[0071] 16 is a functional block diagram of a shoe processing support device 14 according to the third embodiment. The shoe processing support device 14 according to the third embodiment further includes a correction necessity determination unit 112 and a shape correction unit 113. The correction necessity determination unit 112 determines whether correction of the shape of the processing region 65 in the midsole shape data is necessary. The shape correction unit 113 corrects the shape of the processing region 65 corresponding to the bonded region 40 in the midsole shape data based on the shape of the bonded region 40.

[0072] Fig. 17 is a flowchart showing the process S300 for generating processing data for the shoe processing support device 14 of the third embodiment. In the process S300 of Fig. 17, steps S301 to S310 and S313 to S314 are basically the same as steps S101 to S112 of Fig. 3 except for points that are particularly mentioned, and therefore a description thereof will be omitted.

[0073] In step S311, the correction necessity determination unit 112 determines whether correction of the midsole shape data is necessary. FIG. 18 is a diagram for explaining the flow of the shape correction process of this embodiment. FIG. 18 shows the bonded region 40f and the processing region 65f in FIG. 15. As shown in FIG. 18, for example, the correction necessity determination unit 112 calculates counterclockwise vectors Vp and Vq of the edge curves of the bonded region 40f and the processing region 60f, and center of gravity positions Gp and Gq, respectively. For example, when the correction necessity determination unit 112 compares the bonded region 40f and the processing region 60f by superimposing the center of gravity positions Gp and Gq, it identifies a portion in the midsole shape data where the vector Vp differs from Vq by a predetermined amount or more. If a different portion is identified, the correction necessity determination unit 112 identifies the portion as a distorted portion and determines that correction of the midsole shape data is necessary. In the example of FIG. 15, the correction necessity determination unit 112 identifies distorted portions 80f-80h in the midsole shape data.

[0074] In step S312, the shape correction unit 113 corrects the shape of the processing region 65 in the midsole shape data. For example, for the identified distorted portions 80f-80h, the shape correction unit 113 corrects the vector Vq of the distorted portions 80f-80h so that the vector Vp of the portions of the bonded region 40f corresponding to the distorted portions 80f-80h matches the vector Vp of the portions of the bonded region 40f corresponding to the distorted portions 80f-80h, thereby correcting the shape of the processing region 65 in the midsole shape data. FIG. 18 shows a processing region 66f in which the shape of the processing region 65f has been corrected. As shown in FIG. 18, the shape of the corrected portion 81f has been corrected so that it matches the vector Vp of the portions of the bonded region 40f corresponding to the distorted portions 80f-80h.

[0075] If the shape of the processing region 65 in the midsole shape data has been corrected, in steps S313 and S314, processing region data and trajectory data are generated based on the corrected midsole shape data, respectively.

[0076] As shown in the third embodiment, the shape correction unit 113 may correct the shape of the processing area 65 in the midsole shape data that corresponds to the bonded area 40, based on the shape of the bonded area 40. With this configuration, even if the shape of the outline of the processing area 65 in the midsole shape data is distorted due to data noise or the like, it is possible to identify the processing area 65 with an appropriate shape.

[0077] Fourth Embodiment A fourth embodiment of the present disclosure will be described below. In the drawings and description of the fourth embodiment, components and members that are the same as or equivalent to those in the first embodiment will be denoted by the same reference numerals. Descriptions that overlap with the first embodiment will be omitted as appropriate, and the description will focus on configurations that differ from the first embodiment.

[0078] The unit vector of the second moment of area used in the first embodiment is scale invariant and translation invariant, so its value does not change with scale and parallel translation. On the other hand, the unit vector of the second moment of area is rotation dependent, so its value changes depending on the orientation of the midsole shape data and the outsole shape data. Therefore, when the orientations of the midsole shape data and the outsole shape data are misaligned or when the midsole shape data and the outsole shape data are for opposite feet, there is room for improvement in accurately identifying the processing area 65 from the processing area candidates 60.

[0079] Therefore, the identifying unit 107 of this embodiment identifies the candidate processing region 60 for which the deviation between the Hu moment for the candidate processing region 60 and the Hu moment for the welded region 40 is smallest as the processing region corresponding to that welded region 40. The Hu moment is a moment feature that is scale invariant, translation invariant, and rotation invariant. For example, the identifying unit 107 extracts, as the Hu moment, an invariant quantity for translation, scale, and rotation of a certain shape included in the image. Below, general moment feature quantities that are the premise for calculating the Hu moment will be described, and then the Hu moment will be described.

[0080] Nth moment M ｐｑ is a moment about the origin, and is calculated using the following equation (4): In equation (4), F(x, y) is a binary image having pixel values ​​of 1 or 0.

[0081] Here, the center of gravity of the contour in the x direction is x ｃ and the centroid y of the contour in the y direction ｃ are expressed as the following equations (5) and (6), respectively.

[0082] Nth central moment μ ｐｑ is a moment about the center of gravity of the contour, and is calculated using the following equation (7).

[0083] Normalized Nth-order central moment η ｐｑ is the Nth central moment μ ｐｑ is a moment feature normalized by the size of the contour, and is expressed by the following equation (8).

[0084] The Hu moment is a set of multiple Hu element moments h that constitute the Hu moment and have scale invariance, translation invariance, and rotation invariance. ０ ~h ６ Hu element moment h ０ ~h ６ is calculated using the following equations (9) to (15).

[0085] The above is an explanation of the Hu moment. The process for generating processing data for the shoe processing support device 14 of the fourth embodiment will now be described. Figure 19 is a flowchart showing the process S400 for generating processing data for the shoe processing support device 14 of the fourth embodiment. In process S200 of Figure 19, steps S401 to S408 and S411 to S412 are basically the same as steps S101 to S108 and S111 to S112 of Figure 3 except where specifically noted, and therefore a description thereof will be omitted.

[0086] In step S409, the specifying unit 107 calculates the Hu moment h for each welded region 40 and the edge curve of the candidate processing region 60 narrowed down for each welded region 40. Here, the Hu element moment h ６ is an invariant whose sign is reversed for shapes that are reversed left and right, such as a right foot and a left foot, or a mirror image of the shape. Therefore, when the midsole shape data and the outsole shape data are for opposite feet, the Hu element moment h ６ When the Hu moment h is calculated including the above, the Hu element moment h of each of the midsole shape data and the outsole shape data is ６ The sign of the Hu element moment h ０ ~h ６ Of these, Hu element moment h ６ Hu element moment h excluding ０ ~h ５ The Hu moment h is calculated by calculating

[0087] In step S410, the identifying unit 107 identifies the processing region 65 corresponding to the welded region 40 based on the similarity of the shapes of each welded region 40 and each candidate processing region 60. For example, the identifying unit 107 identifies the candidate processing region 60 having the smallest deviation between the Hu moment h of the candidate processing region 60 and the Hu moment h of the welded region 40 as the processing region corresponding to the welded region 40. Here, the Hu element moment h in the Hu moment h ０ ~h ５ Therefore, the scale of the Hu element moment h, which has a particularly large scale, ０ and h １ Hu element moment h other than ２ ~h ５ In order to take into account the above, the specifying unit 107 of this embodiment calculates the Hu element moment h ０ ~h ５ Based on the maximum and minimum values ​​of the Hu element moment h of each of the candidate processing area 60 and the welded area 40, ０ ~h ５ Normalized Hu element moment hu by scaling and normalizing ０ ~hu ０ and generate the normalized Hu element moment hu ０ ~hu ５ The processing region is identified using the Hu element moment h ０ ~h ５ Based on the above maximum and minimum values, the Hu element moments h of both the candidate processing area 60 and the welded area 40 are calculated. ０ ~h ５ Here, the normalized Hu moment for the candidate processing region 60 is expressed as hu pro and the normalized Hu moment hu pro The k-th normalized Hu element moment hu pro,k (k=0, 1, 2, 3, 4, or 5). The normalized Hu moment for the bonded region 40 is defined as hu bond and the normalized Hu moment hu bond The k-th normalized Hu element moment Hu bond,kFor example, the specifying unit 107 calculates the normalized Hu moment hu of each candidate 60 for the processing region. pro and the normalized Hu moment hu of each bonded region 40 bond Based on this, the Euclidean distance d between each welded region 40 and the candidate processing region 60 narrowed down for each welded region 40 is calculated. The Euclidean distance d here is an example of "the deviation between the Hu moment h of the candidate processing region 60 and the Hu moment h of the welded region 40." The Euclidean distance d in the fourth embodiment is calculated using the following formula (16).

[0088] If the Hu element moment h ０ ~h ５ Without normalizing, the Hu element moment h ０ ~h ５ When calculating the Euclidean distance using ０ and h １ The value of the Euclidean distance is almost determined by the elements calculated in the above. ０ ~h ５ By normalizing and scaling, the lower-order Hu element moments h ０ and h １ Higher order Hu element moments h ２ ~h ５ The elements calculated in can also contribute to the calculation of the Euclidean distance, allowing for more detailed evaluation of similarity.

[0089] The identifying unit 107 identifies the candidate processing area 60 with the smallest calculated Euclidean distance as the processing area 65 corresponding to the to-be-welded area 40 .

[0090] In step S410, the identification unit 107 may identify the processing area 65 corresponding to the to-be-bonded area 40 using another method, such as a distance calculation method used in the image matching process in OpenCV. For example, according to OpenCV, the processing area 65 corresponding to the to-be-bonded area 40 may be identified using any of the following formulas (17) to (19). In this case, the candidate processing area 60 having the smallest value M calculated using any of formulas (17) to (19) may be identified as the processing area corresponding to the to-be-bonded area 40.

[0091] Here, m in formulas (17) to (19) bond,k and m pro,k is calculated using the following equations (20) and (21).

[0092] After steps S411 and S412, the process S400 ends.

[0093] FIG. 20 shows the accuracy rate of matching when the processing area 65 is identified using each identification method. FIG. 21 illustrates the midsole shape data and outsole shape data used in FIG. 20. FIG. 21(a) shows an example of the joint surface 6a of the outsole 6 in the outsole shape data. FIG. 21(b) shows an example of the joint surface 8a of the midsole 8 in the midsole shape data arranged in the same orientation as the joint surface 6a in FIG. 21(a). FIG. 21(c) shows an example of the joint surface 8a of the midsole 8 in the midsole shape data for a shoe on the opposite foot from the shoe used in FIG. 21(a). FIG. 21(d) shows an example of the joint surface 8a of the midsole 8 in the midsole shape data arranged at an angle of 10° relative to the joint surface 6a in FIG. 21(a). FIG. 21(e) shows an example of the joint surface 8a of the midsole 8 in the midsole shape data arranged at an angle of 20° relative to the joint surface 6a in FIG. 21(a). In FIG. 20, the vertical axis indicates the accuracy rate of the result of specifying the processing area, and legends E1 to E4 indicate the examples of FIGS. 21(b) to 21(e), respectively.

[0094] Case C1 shows the case where the processing area is identified using the second moment of area converted into a unit vector (i.e., the case of the first embodiment). In case C1, the accuracy rate is relatively low in legends E2 and E4. That is, it can be seen that the accuracy rate decreases in the case of the opposite foot and when the direction is changed significantly.

[0095] In case C2, the Hu element moment h before normalization ０ ~h ６ This shows a case where the processing area is identified using the Euclidean distance. In case C2, the accuracy rate for legend E2 is relatively low. In other words, it can be seen that the accuracy rate decreases in the case of the opposite foot. Therefore, it can be seen that the decrease in accuracy rate is suppressed when the direction changes.

[0096] Case C3 is the normalized Hu element moment hu ０ ~hu ６ 1 shows a case where the processing area is identified using the Euclidean distance. In case C3, the accuracy rate for legend E2 is relatively low, but it is higher than case C2. In other words, it can be seen that the accuracy rate for the opposite foot case can be improved.

[0097] Case C4 is the normalized Hu element moment hu ０ ~hu ５ Using the Euclidean distance for ６ This shows the case where the processing area is specified excluding the legends E1 to E4 (i.e., the case of the second embodiment). In case C4, a high accuracy rate is obtained for all legends E1 to E4. In other words, it can be seen that the decrease in accuracy rate when the direction is different or when the foot is the opposite is significantly suppressed.

[0098] Case C5 is the normalized Hu element moment hu ０ ~hu ５ This shows a case where the processing area is identified using the value M calculated using equation (17). In case C5, a high accuracy rate is obtained for all legends E1 to E4. In case C1, the accuracy rate for legend E4 is relatively low. In other words, it can be seen that the accuracy rate decreases when the orientation is changed significantly.

[0099] As shown in the fourth embodiment, the identification unit 107 executes a first step of calculating the Hu moment for the candidate processing area 60 and the Hu moment for the bonded area 40, and a second step of identifying the candidate processing area 60 for which the difference between the Hu moment calculated for the candidate processing area 60 and the Hu moment calculated for the bonded area 40 is smallest as the processing area 65 corresponding to that bonded area 40. With this configuration, it is possible to accurately identify the processing area 65 from the candidate processing area 60 even if the orientations of the midsole shape data and the outsole shape data are misaligned or for the opposite foot.

[0100] As shown in the fourth embodiment, the first step calculates a plurality of Hu element moments that constitute Hu moments and have scale invariance, translation invariance, and rotation invariance for the candidate processing region 60 and the welded region 40, thereby calculating the Hu moments for the candidate processing region 60 and the welded region 40; the second step includes a step of normalizing the Hu element moments calculated for the candidate processing region 60 and the Hu element moments calculated for the welded region 40 based on the maximum and minimum values ​​of the plurality of Hu element moments to calculate normalized Hu element moments for the candidate processing region 60 and the welded region 40; and a step of calculating the deviation based on the normalized Hu element moments calculated for the candidate processing region 60 and the normalized Hu element moments calculated for the welded region 40. This configuration makes it possible to calculate the deviation by taking into account not only low-order Hu element moments but also high-order Hu element moments. As a result, it becomes possible to evaluate the similarity between the candidate processing area 60 and the to-be-bonded area 40 in more detail, and it becomes possible to specify the processing area 65 with high accuracy.

[0101] As described in the fourth embodiment, the specifying unit 107 specifies the Hu element moment h ６ (i.e., Hu element moment h０ ~h ５ ) to calculate the Hu moment for the candidate processing region 60 and the Hu moment for the welded region 40. Here, assuming that the Hu element moment h ６ When using the Hu element moment h ０ ~h ６ When normalized, the Hu element moment h ６ Therefore, when the midsole shape data and the outsole shape data are for opposite feet, the work area 65 may not be accurately identified. ６ Since the above formula is not used, it is possible to accurately identify the processing area 65 even if the midsole shape data and the outsole shape data are for opposite feet.

[0102] Fifth Embodiment A fifth embodiment of the present disclosure will be described below. In the drawings and description of the fifth embodiment, components and members that are the same as or equivalent to those in the first embodiment will be denoted by the same reference numerals. Explanations that overlap with the first embodiment will be omitted as appropriate, and the description will focus on configurations that differ from the first embodiment.

[0103] The shoe processing support device 14 of the first to fourth embodiments extracts the to-be-joined region 40 based on the outsole shape data. However, there may be cases where the outsole shape data of the outsole 6 to be joined is not stored in advance in the storage unit 101, or where the quality of the outsole shape data is so poor that the to-be-joined region 40 cannot be extracted. Therefore, a method for identifying the processing region 65 without requiring the outsole shape data of the outsole 6 to be joined is required. Based on this, the fifth embodiment will be described below.

[0104] FIG. 22 is a functional block diagram of a shoe processing support device 14 according to a fifth embodiment. The storage unit 101 according to the fifth embodiment stores a prediction model 114 that has been machine-learned using data including feature quantities for the processing areas 65 of shoe components as learning data. The "data including feature quantities for the processing areas 65 of shoe components" here may be image data of shoe components, individual processing areas 65 of shoe components, a combination of multiple shoe components, or a shoe composed of multiple shoe components, or may be feature quantities extracted from such image data. The "feature quantities for the processing areas 65" may directly indicate feature quantities for the processing areas 65, such as the feature quantities of the processing areas 65 themselves, or may indirectly indicate feature quantities for the processing areas 65, such as the feature quantities of the bonded areas 40 to be bonded to the processing areas 65. In this embodiment, the feature quantities for the processing areas 65 are feature quantities for the bonded areas 40 to be bonded to the processing areas 65. It should be noted that in the fifth embodiment, the to-be-joined region 40 may refer not only to the to-be-joined region 40 of the outsole 6 to be joined, but also to the to-be-joined region 40 included in any outsole 6 (for example, an outsole 6 that is not to be joined (other than the outsole 6 to be joined)). Similarly, it should be noted that the processed region 65 may refer not only to the to-be-joined region 65 of the midsole 8 to be joined, but also to the processed region 65 included in any midsole 8 (for example, a midsole 8 that is not to be joined (other than the midsole 8 to be joined)).

[0105] The shoe processing support device 14 of the fifth embodiment further includes a feature extraction unit 115 that extracts feature amounts for the candidate processing area 60. In this embodiment, the feature amounts for the candidate processing area 60 include shape feature amounts relating to the shape of the candidate processing area 60 and position feature amounts relating to the position of the candidate processing area 60. The shape feature amounts include, for example, moment feature amounts such as the second moment of area and Hu moment, the relative area of ​​the area with respect to the entire shoe sole, and the number of parts. The position feature amount includes, for example, a position vector of the normalized center of gravity position.

[0106] The position feature may also be the ranking of the center of gravity position (e.g., ranking of proximity to the toe, ranking of proximity to the medial side of the foot, ranking of proximity to the heel) or the distance from other candidate processing areas 65. For example, since the outsole 6 is located in an area that is likely to come into contact with the ground, and the heel and toe are covered by the outsole 6 in most shoes, it is considered that there is a tendency for the processing area 65 to be more likely to be located in a position close to the heel or close to the toe. Furthermore, if only relative position feature such as the shape feature and the position vector of the center of gravity position is used, it is expected that the candidate processing area 60 near the midfoot, where the centers of gravity of the processing areas 65 tend to be concentrated, will be determined as the processing area 65, and the candidate processing area 60 near the end will not be determined as the processing area 65. Therefore, the accuracy of determining the processing area 65 can be improved by adding the ranking of the center of gravity position, which is an index representing the positional relationship of each area, and the distance from other candidate processing areas 65 as the position feature and weighting the score according to the ranking and distance.

[0107] The position feature amount may also be the sum of existence probabilities. For example, the existence probability may be calculated by filling the outline of the candidate processing area 65 with a grid of points and counting how many processing areas in the learning data the point is in. The existence probability of the area may be determined by adding up all the existence probabilities of each point in the candidate processing area 65 or by dividing the sum by the number of points in the candidate processing area 65.

[0108] The feature extraction method is not limited to the above example, and the feature extraction unit 115 may extract features using known feature extraction techniques such as SIFT (Scale-Invariant Feature Transform), SURF (Speed-Upped Robust Feature), AKAZE (Accelerated-KAZE), etc. Alternatively, the feature extraction unit 115 may extract features using a convolutional neural network.

[0109] The shoe processing support device 14 of the fifth embodiment includes, instead of the identification unit 107, an area determination unit 116 that determines whether the candidate processing area 60 is the processing area 65 from the feature amounts extracted by the feature amount extraction unit 115 using a prediction model 114. The area determination unit 116 of this embodiment calculates the likelihood that the candidate processing area 60 is the processing area 65 from the feature amounts extracted using the prediction model, and identifies the processing area 65 from the candidate processing area 60 based on the calculated likelihood. The area determination unit 116 of this embodiment determines that the candidate processing area 60 is the processing area 65 when, for example, the calculated likelihood exceeds a threshold. The "likelihood that the candidate processing area 60 is the processing area 65" can also be interpreted as, for example, the likelihood that the determination result or identification result for the processing area 65 is correct (the probability that the determination result is correct), or the degree of certainty of the determination result or identification result.

[0110] The shoe processing assisting device 14 of the fifth embodiment further includes a learning processing unit 117 that executes machine learning processing to generate a predictive model 114 and stores the generated predictive model 114 in the storage unit 101. The learning processing unit 117 of the present embodiment uses a regression model such as a random forest or a neural network as a machine learning algorithm.

[0111] The learning process of the prediction model 114 will now be described. First, in order to construct the prediction model 114, a photograph of the sole of a shoe is taken in advance to obtain a captured image. From the captured image, regions surrounded by at least one of the outer contour of the sole and the ribs, whose area is equal to or greater than a predetermined area, are extracted as candidates for the bonded regions 40 of the outsole 6 to be learned. Here, known or future technology may be used to extract the regions surrounded by at least one of the outer contour shape of the bonded regions 40 of the outsole 6 and the ribs. Next, the feature extraction unit 115 extracts feature amounts of the candidate bonded regions 40 of the outsole 6 in the captured image. Next, for each candidate bonded region 40, an operator inputs, via an input device (not shown), a correct label indicating that the candidate bonded region 40 is a bonded region 40 and an incorrect label indicating that the candidate bonded region 40 is not a bonded region 40. Next, the learning processing unit 117 performs machine learning using pairs of feature amounts and correct labels of the candidate bonded regions 40 and pairs of feature amounts and incorrect labels of the candidate bonded regions 40 as learning data. As a result, a prediction model 114 is created and stored in the storage unit 101. In the prediction model 114 of this embodiment, when the shape feature values ​​and position feature values ​​for the candidate machining area 60 extracted by the feature value extraction unit 115 are input, the prediction model 114 outputs the likelihood that the candidate machining area 60 is the machining area 65. The closer the feature values ​​of the candidate machining area 60 are to the feature values ​​for the welded area 40, the larger the value of this "likelihood that the candidate machining area is the machining area".

[0112] Fig. 23 is a flowchart showing the process S500 for generating processing data for the shoe processing support device 14 of the fifth embodiment. In the process S500 of Fig. 23, steps S501 to S503 and S507 to S508 are basically the same as steps S101 to S103 and S111 to S112 of Fig. 3 except where specifically mentioned, and therefore a description thereof will be omitted.

[0113] In step S504, the feature extraction unit 115 extracts the feature of the candidate processing area 60.

[0114] In step S505 , the area determination unit 116 inputs the extracted feature amount into the prediction model 114 to calculate the likelihood that the candidate processing area 60 is the processing area 65 .

[0115] In step S506, the area determination unit 116 determines whether the candidate for machining area 60 is the machining area 65 based on the calculated likelihood that the candidate for machining area 60 is the machining area 65. The area determination unit of this embodiment determines that the candidate for machining area 60 is the machining area 65 when the calculated likelihood that the candidate for machining area 60 is the machining area 65 is greater than a predetermined threshold.

[0116] 24 illustrates the likelihood that regions 60a to 60e, which are candidate regions 60 for the processing region, are processing regions 65. In the example of FIG. 24, the likelihoods calculated for regions 60a to 60e are 0.91, 0.99, 0.90, 0.95, and 0.65, respectively. The region determination unit 116 compares the likelihood for each region 60a to 60e with a threshold value (e.g., 0.85). The region determination unit 116 determines that regions 60a to 60d with a likelihood greater than the threshold value are processing regions 65, and determines that regions 60a to 60d with a likelihood equal to or less than the threshold value are not processing regions 65.

[0117] Thereafter, steps S507 and S508 are executed, and the process S500 ends.

[0118] As shown in the fifth embodiment, the shoe processing support device 14 includes a first acquisition unit 102 that acquires imaging data of the joining surface of the first shoe component, a shape data generation unit 103 that generates first shape data indicating the shape of the joining surface of the first shoe component based on the imaging data, a first extraction unit 104 that extracts, based on the first shape data, candidate processing areas 60 to be processed for joining to a joining surface of a second shoe component that is joined to the first shoe component by the joining surface, a feature extraction unit 115 that extracts feature amounts of the candidate processing areas 60, and an area determination unit 116 that determines whether the candidate processing area 60 is a processing area 65 from the extracted feature amounts using a prediction model 114 that has been machine-learned using data including the feature amounts of the processing areas of the shoe component as learning data. This configuration makes it possible to accurately determine the processing area 65 without requiring outsole shape data.

[0119] As shown in the fifth embodiment, the area determination unit 116 uses the prediction model 114 to calculate the likelihood that the candidate processing area 60 is the processing area 65 from the extracted feature amount. With this configuration, it is possible to more accurately determine that the candidate processing area 60 is the processing area 65 without requiring outsole shape data.

[0120] As shown in the fifth embodiment, when the calculated likelihood exceeds a threshold, the area determination unit 116 determines that the candidate processing area 60 is the processing area 65. According to this configuration, it is possible to more accurately determine that the candidate processing area 60 is the processing area 65 without requiring outsole shape data.

[0121] As shown in the fifth embodiment, the feature amounts of the processing area candidate 60 include shape feature amounts relating to the shape of the processing area candidate 60 and position feature amounts relating to the position of the processing area candidate 60. With this configuration, it is possible to determine with higher accuracy that the processing area candidate 60 is the processing area 65, without requiring outsole shape data, compared to when the feature amounts of the processing area candidate 60 include only one of shape feature amounts and position feature amounts.

[0122] As shown in the fifth embodiment, the learning data includes a pair of feature amounts for the processing area 65 and a correct label indicating that the area is the processing area 65, and a pair of feature amounts for an area that is not the processing area 65 and an incorrect label indicating that the area is not the processing area 65. With this configuration, it becomes possible to more accurately determine that the processing area candidate 60 is the processing area 65 without requiring outsole shape data.

[0123] In the fifth embodiment, a regression model is used as the machine learning algorithm in the prediction model 114, but this is not limited to this, and a distance learning model may also be used. In this case, for example, the prediction model 114 may be created by performing supervised learning to learn the classification boundary between the bonded region 40 and the region that is not the bonded region 40 using the feature amount of the bonded region 40 and the feature amount of the region that is not the bonded region 40. Then, the processing region 65 may be determined by using this prediction model 114 to evaluate which side of the classification boundary between the bonded region 40 and the region that is not the bonded region 40 the feature amount of the candidate processing region 65 belongs to.

[0124] Furthermore, when a distance learning model is used as the machine learning algorithm in the prediction model 114, the prediction model 114 may be created by performing unsupervised learning using the feature amounts of the welded region 40. Then, using this prediction model 114, when the distance between the feature amount of the candidate processing region 60 and the feature amount of its nearest neighbor does not exceed a threshold, the candidate processing region 60 may be determined to be the processing region 65.

[0125] Furthermore, a regression model and a metric learning model may be combined as the machine learning algorithm in the prediction model 114. In this case, the average (or weighted value) of the scores calculated from the two models, the regression model and the metric learning model, may be used.

[0126] In the fifth embodiment, the processing area 65 is identified by comparing the degree of similarity of the feature amount with a threshold value, but this is not limiting. For example, the area determination unit 116 may determine that a specified number of candidate processing areas 60 are the processing areas 65 in descending order of the degree of similarity of the feature amount. In this case, the number of candidate processing areas 60 may be specified by a user input to a user input device (not shown). For example, if the user specifies the number of processing areas 65 as four, the top four candidate processing areas 60 with the greatest degree of similarity of the feature amount may be determined to be the processing areas 65. This configuration makes it possible to more accurately determine that the candidate processing areas 60 are the processing areas 65 without requiring outsole shape data.

[0127] In the fifth embodiment, the prediction model 114 is created by learning the feature amounts of the bonded region 40 in the outsole 6, but this is not limited to this, and the prediction model 114 may be created by learning the feature amounts of the processing region 65 in the midsole 8. In this case, the "likelihood that the candidate processing region is the processing region" output from the prediction model 114 becomes larger the closer the feature amounts of the candidate processing region 60 are to the feature amounts of the processing region 65. Furthermore, the prediction model 114 may be created by learning both the feature amounts of the bonded region 40 in the outsole 6 and the feature amounts of the processing region 65 in the midsole 8.

[0128] A predictive model 114 may be created for each position or region. For example, a predictive model 114 may be created for each of the toe region, heel region, and regions therebetween. In this case, different weights may be applied to each predictive model 114 created for each position or region.

[0129] In the fifth embodiment, the learning processing unit 117 is provided in the shoe processing support device 14, but is not limited to this and may be provided in a device (such as a server) external to the shoe processing support device 14. In this case, the area determination unit 116 may determine the processing area 65 using the prediction model 114 generated by the learning processing unit 117 in the external device.

[0130] A modification of the above embodiment will now be described.

[0131] In the above embodiment, the bonded region 40 is extracted based on the outsole shape data stored in advance in the storage unit 101, but this is not limiting. For example, the bonded region 40 may be extracted based on outsole shape data obtained by imaging the outsole 6. With this configuration, it is possible to accurately identify the processing region 65 based on the bonded region 40 extracted from the shape of the bonded surface 6a of the outsole 6 that will actually be bonded.

[0132] In the above embodiment, the outsole 6 and the midsole 8 are joined by bonding them together, but this is not limiting. For example, the outsole 6 and the midsole 8 may be joined together by sewing them together. In this case, the "processing" may be, for example, a process of making a hole for passing a needle through.

[0133] In the above embodiment, candidates for the processing area are extracted based on the shape of the rib 50, but this is not limiting. For example, the processing area may be identified based on the shape of at least one of the outer contours of a concave portion and a convex portion, such as a spike-like protrusion, a grip portion, or a depression, formed on the midsole 8.

[0134] The method of specifying the center of gravity position by the specifying unit 107 and the method of narrowing down the candidates for the processing area 60 are not limited to the above-mentioned example. Fig. 25 is a diagram for explaining another example of the method of specifying the center of gravity position by the specifying unit 107 and the candidates for the processing area 60. Fig. 25 shows the joining area 40 of the outsole 6. １ ~40 ８ As shown in FIG. 25, for example, the specifying unit 107 creates a boundary frame 80 that surrounds the entire contour of the to-be-welded region 40, and defines the coordinates of the four vertices of the boundary frame 80 as [x ｍｉｎ , y ｍｉｎ ]=[0,0],[x ｍａｘ , y ｍｉｎ ]=[1,0],[x ｍｉｎ , y ｍａｘ ]=[0,1],[x ｍａｘ , y ｍａｘ ]=[1, 1], and then, １ ~40 ８ Center of gravity position of the contour GO１ ~GO ８ Here, each of the bonded regions 40 １ ~40 ８ Center of gravity position of the contour GO １ ~GO ８ The coordinates of [xg ｉ , yg ｉ ] (where i is a natural number equal to or less than 8). In the example of FIG. 25, each of the bonded regions 40 １ ~40 ８ Center of gravity position of the contour GO １ ~GO ８ Coordinates of [x ｉ , y ｉ ] are calculated as [0.57, 0.08], [0.26, 0.28], [0.81, 0.22], [0.80, 0.34], [0.27, 0.72], [0.84, 0.63], [0.85, 0.75], and [0.56, 0.91], respectively. The normalized center of gravity coordinates are [x norm,ｉ , y norm,ｉ ], the specifying unit 107 uses the following formula (22) to determine the distance between each bonded region 40 １ ~40 ８ The normalized coordinates of the center of gravity of the contour [x norm,ｉ , y norm,ｉ ] can be calculated.

[0135] Similarly, the specifying unit 107 calculates the normalized center of gravity coordinates [x norm,ｉ , y norm,ｉ The specifying unit 107 calculates the distance between the bonded regions 40 １ ~40 ８ and the normalized center of gravity coordinates [x norm,ｉ , y norm,ｉ ] to calculate the Euclidean distance d, thereby narrowing down the candidates 65 for the processing area.

[0136] In the first to fourth embodiments, the candidate processing region 60 corresponding to the welded region 40 is narrowed down, but this is not limiting. For example, the processing region 65 corresponding to the welded region 40 may be identified by calculating the deviation between the moment feature amount for the candidate processing region 60 and the moment feature amount for the welded region 40 in a brute force manner, without narrowing down the candidate processing region 60 corresponding to the welded region 40.

[0137] In the first to fourth embodiments, the Euclidean distance is used, but this is not limiting. For example, the Minkowski distance may be used instead of the Euclidean distance.

[0138] The present disclosure has been described above based on the embodiments. These embodiments are merely examples, and it will be understood by those skilled in the art that various modifications are possible in the combination of the respective components and processing steps, and that such modifications are also within the scope of the present disclosure. Furthermore, the above-described embodiments can be generalized to provide the following aspects.

[0139] A shoe processing support method comprising: a step of acquiring image data of a joining surface of a first shoe component; a step of generating first shape data indicating the shape of the joining surface of the first shoe component based on the image data; a step of extracting, based on the first shape data, candidate processing areas to be processed for joining to a joining surface of a second shoe component that is to be joined to the first shoe component by the joining surface; a step of acquiring second shape data indicating the shape of the joining surface of the second shoe component; a step of extracting, based on the second shape data, the joining area that is to be joined to the first shoe component by the joining surface; and a step of identifying the processing areas among the candidate processing areas that correspond to each of the joining areas based on a comparison between the candidate processing areas and the joining areas.

[0140] This allows the processing area on the joint surface of the first shoe component to be identified from the image data of the joint surface of the first shoe component, making it possible to perform appropriate processing on the joint surface of the first shoe component depending on manufacturing errors, etc.

[0141] [Aspect 2] The shoe processing support method according to aspect 1, further comprising the step of generating processing area data for each of the identified processing areas, the processing area having an outer periphery offset inward by a predetermined distance from the outer periphery of the processing area.

[0142] This prevents the processing from going beyond the outer periphery of the processing area.

[0143] [Aspect 3] The shoe processing support method according to aspect 2, further comprising the step of generating trajectory data indicating a trajectory when processing the processing area based on the processing area data.

[0144] This prevents the processing from going beyond the outer periphery of the processing area.

[0145] [Aspect 4] The shoe processing support method according to any one of Aspects 1 to 3, further comprising the step of correcting the shape of the processing area corresponding to the to-be-welded area in the first shape data based on the shape of the to-be-welded area.

[0146] This makes it possible to identify the processing area with an appropriate shape even if the shape of the outline of the processing area in the midsole shape data is distorted due to data noise or the like.

[0147] [Aspect 5] The shoe processing support method according to any one of Aspects 1 to 4, wherein the step of extracting the candidate processing area extracts the candidate processing area based on the shape of the outline of the joint surface of the first shoe component and the shape of the outline of at least one of the recessed portion and the protruding portion formed on the joint surface of the first shoe component.

[0148] This makes it possible to appropriately extract candidates for the processing region.

[0149] [Aspect 6] The shoe processing support method according to any one of Aspects 1 to 5, wherein the step of extracting the bonded region extracts the bonded region based on the second shape data stored in advance in a storage unit.

[0150] This eliminates the need to actually photograph the joining surface of the second shoe component in the process for generating processing data for the shoe processing support device described above, thereby contributing to improved work efficiency and reduced work costs.

[0151] [Aspect 7] The shoe processing support method according to Aspect 6, wherein the step of extracting the bonded area includes a step of correcting the size of the bonded area in the second shape data based on the size of the bonded surface of the first shoe component and the size of the bonded surface of the second shoe component stored in advance in the storage unit.

[0152] This allows the processing area to be appropriately identified even when the first shoe component and the second shoe component are different in size. Therefore, it is only necessary to register one piece of second shape data for each product type, which makes it possible to suppress an increase in the amount of data.

[0153] [Aspect 8] The shoe processing support method according to any one of Aspects 1 to 5, wherein the step of extracting the welded area extracts the welded area based on the second shape data obtained by imaging the second shoe component.

[0154] This makes it possible to accurately identify the processing area based on the to-be-joined area extracted from the shape of the to-be-joined surface of the second shoe component that is actually to be joined.

[0155] [Aspect 9] The shoe processing support method according to any one of aspects 1 to 8, further comprising a step of determining whether the shape of the joining surface of the first shoe component is acceptable or not based on a comparison of the shape of the processing area and the shape of the joining area.

[0156] This makes it possible to select a first shoe component having a joining surface with an appropriate shape with high accuracy.

[0157] [Aspect 10] The shoe processing support method according to Aspect 9, wherein the step of determining whether the shape of the joining surface of the first shoe component is acceptable is determined when a deviation between each of the second moments of area in the foot length direction and the foot width direction of the processing area and each of the second moments of area in the foot length direction and the foot width direction of the joined area corresponding to the processing area is equal to or less than a threshold value.

[0158] This makes it possible to appropriately determine whether the shape of the joining surface of the first shoe component is acceptable or not.

[0159] [Aspect 11] The shoe processing support method according to any one of Aspects 1 to 10, wherein the step of specifying the processing area specifies the candidate processing area having the smallest deviation between the second moments of area in each of the foot length direction and the foot width direction of the candidate processing area and the second moments of area in each of the foot length direction and the foot width direction of the bonded area as the processing area corresponding to the bonded area.

[0160] This makes it possible to appropriately identify the processing region corresponding to the region to be welded.

[0161] Aspect 12: The shoe processing support method according to any one of Aspects 1 to 10, wherein the step of specifying the processing area includes a first step of calculating the Hu moment for the candidate processing area and the Hu moment for the bonded area, and a second step of specifying the candidate processing area for which the Hu moment calculated for the candidate processing area and the Hu moment calculated for the bonded area have the smallest difference as the processing area corresponding to that bonded area.

[0162] This makes it possible to accurately identify the processing area 65 from the processing area candidates 60 even if the midsole shape data and outsole shape data are misoriented or for the opposite foot.

[0163] Aspect 13: The shoe processing support method according to Aspect 12, wherein the first step calculates a Hu moment for the candidate processing area and a Hu moment for the joined area by calculating a plurality of Hu element moments that constitute the Hu moment and that are scale invariant, translation invariant, and rotation invariant for the candidate processing area and the joined area; and the second step includes a step of normalizing the Hu element moment calculated for the candidate processing area and the Hu element moment calculated for the joined area based on maximum and minimum values ​​of the plurality of Hu element moments to calculate a normalized Hu element moment for the candidate processing area and a normalized Hu element moment for the joined area; and a step of calculating the deviation based on the normalized Hu element moment calculated for the candidate processing area and the normalized Hu element moment calculated for the joined area.

[0164] This makes it possible to calculate the deviation by taking into consideration not only the low-order Hu element moments but also the high-order Hu element moments. As a result, it becomes possible to evaluate the similarity between the candidate processing region 60 and the welded region 40 in more detail, and therefore it becomes possible to identify the processing region 65 with high accuracy.

[0165] [Aspect 14] The shoe processing support device according to Aspect 13, wherein the first step calculates the Hu moment for the candidate processing area and the Hu moment for the bonded area without using any Hu element moment, which is an invariant that has the property of being inverted in sign with respect to a mirror image of the Hu moment, among the plurality of Hu element moments.

[0166] This makes it possible to accurately identify the processing area 65 even if the midsole shape data and the outsole shape data are for opposite feet.

[0167] A shoe processing support device comprising: a first acquisition unit that acquires imaging data of a joining surface of a first shoe component; a shape data generation unit that generates first shape data indicating a shape of the joining surface of the first shoe component based on the imaging data; a first extraction unit that extracts, based on the first shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is to be joined to the first shoe component by the joining surface; a second acquisition unit that acquires second shape data indicating the shape of the joining surface of the second shoe component; a second extraction unit that extracts, based on the second shape data, the joining area that is to be joined to the first shoe component by the joining surface; and an identification unit that identifies the processing areas of the candidate processing areas that correspond to each of the joining areas based on a comparison between the candidate processing areas and the joining areas.

[0168] This makes it possible to obtain the same effect as the shoe processing support method of aspect 1.

[0169] A shoe processing support program that causes a computer to execute the following steps: acquiring image data of a joining surface of a first shoe component; generating first shape data based on the image data, the first shape data indicating the shape of the joining surface of the first shoe component; extracting, based on the first shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is to be joined to the first shoe component by the joining surface; acquiring second shape data indicating the shape of the joining surface of the second shoe component; extracting, based on the second shape data, the joining area that is to be joined to the first shoe component by the joining surface; and identifying, from the candidate processing areas, the processing areas that correspond to each of the joining areas based on a comparison between the candidate processing areas and the joining areas.

[0170] This makes it possible to obtain the same effect as the shoe processing support method of aspect 1.

[0171] [Aspect 17] A shoe processing support method comprising: a step of acquiring imaging data of a joining surface of a first shoe component; a step of generating shape data indicating the shape of the joining surface of the first shoe component based on the imaging data; a step of extracting, based on the shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is joined to the first shoe component by the joining surface; a step of extracting feature values ​​of the candidate processing areas; and a step of determining whether the candidate processing area is the processing area from the extracted feature values ​​using a prediction model that has been machine-learned using data including the feature values ​​for the processing area of ​​the shoe component as training data.

[0172] This makes it possible to determine the processing area 65 with high accuracy without requiring outsole shape data.

[0173] [Aspect 18] The shoe processing support method according to aspect 17, wherein the determining step includes a step of calculating, using the prediction model, a likelihood that the candidate processing area is the processing area from the extracted feature amount.

[0174] This makes it possible to determine with higher accuracy whether the candidate processing area 60 is the processing area 65.

[0175] [Aspect 19] The shoe processing support method according to aspect 18, wherein the determining step includes a step of determining that the candidate processing area is the processing area when the calculated likelihood exceeds a threshold value.

[0176] This makes it possible to determine with higher accuracy whether the candidate processing area 60 is the processing area 65.

[0177] [Aspect 20] The shoe processing support method according to aspect 18, wherein the determining step determines a designated number of candidates for processing areas to be processing areas in descending order of the calculated likelihood.

[0178] This makes it possible to determine with higher accuracy whether the candidate processing area 60 is the processing area 65.

[0179] [Aspect 21] The shoe processing support method according to any one of Aspects 17 to 20, wherein the feature amounts of the candidate processing area include a shape feature amount relating to a shape of the candidate processing area and a position feature amount relating to a position of the candidate processing area.

[0180] This makes it possible to determine with greater accuracy that the candidate for processing area 60 is the processing area 65 than when the feature amount of the candidate for processing area 60 includes only one of the shape feature amount and the position feature amount.

[0181] [Aspect 22] The shoe processing support method according to any one of Aspects 17 to 21, wherein the learning data includes a pair of feature amounts for the processing area and a correct label indicating that the area is the processing area, and a pair of feature amounts for an area that is not the processing area and an incorrect label indicating that the area is not the processing area.

[0182] This makes it possible to determine with higher accuracy whether the candidate processing area 60 is the processing area 65.

[0183] [Aspect 23] A shoe processing support device comprising: an acquisition unit that acquires imaging data of a joining surface of a first shoe component; a shape data generation unit that generates shape data indicating the shape of the joining surface of the first shoe component based on the imaging data; a candidate extraction unit that extracts candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is joined to the first shoe component by the joining surface, based on the shape data; a feature extraction unit that extracts features of the candidate processing areas; a storage unit that stores a prediction model that has been machine-learned using data including the features of the processing areas of the shoe components as learning data; and a determination unit that uses the prediction model to determine whether the candidate processing area is the processing area from the extracted features.

[0184] This makes it possible to obtain the same effect as the shoe processing support method of aspect 17.

[0185] [Aspect 24] A shoe processing support program that causes a computer to execute the following steps: acquiring image data of a joining surface of a first shoe component; generating shape data indicating the shape of the joining surface of the first shoe component based on the image data; extracting, based on the shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component that is joined to the first shoe component by the joining surface; extracting feature values ​​of the candidate processing areas; and determining, from the extracted feature values, whether the candidate processing area is the processing area, using a prediction model that has been machine-trained using data including the feature values ​​for the processing area of ​​the shoe component as learning data.

[0186] This makes it possible to obtain the same effect as the shoe processing support method of aspect 17.

[0187] The present disclosure relates to a shoe processing support method, a shoe processing support device, and a shoe processing support program.

[0188] 1 Shoe processing system, 6 Outsole, 8 Midsole, 10 Camera, 14 Shoe processing support device, 16 Buffing device, 18 Coating device, 101 Memory unit, 102 First acquisition unit, 103 Shape data generation unit, 104 First extraction unit, 105 Second acquisition unit, 106 Second extraction unit, 107 Identification unit, 108 Processed data generation unit, 109 Buffing processing control unit, 110 Coating control unit, 111 Pass / fail determination unit, 112 Correction necessity determination unit, 113 Shape correction unit, 114 Prediction model, 115 Feature extraction unit, 116 Area determination unit, 117 Learning processing unit.

Claims

1. A shoe processing support method comprising: a step of acquiring image data of a joining surface of a first shoe component; a step of generating first shape data indicative of a shape of the joining surface of the first shoe component based on the image data; a step of extracting, based on the first shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component to be joined to the first shoe component by the joining surface; a step of acquiring second shape data indicative of a shape of the joining surface of the second shoe component; a step of extracting, based on the second shape data, a joining area to be joined to the first shoe component by the joining surface; and a step of identifying the processing areas corresponding to each of the joining areas among the candidate processing areas based on a comparison between the candidate processing areas and the joining areas.

2. The shoe processing support method of claim 1, further comprising a step of generating processing area data indicating, for each of the identified processing areas, a processing area having an outer periphery offset inwardly by a predetermined distance from the outer periphery of the processing area.

3. The shoe processing support method according to claim 2, further comprising a step of generating trajectory data indicating a trajectory when processing the processing area based on the processing area data.

4. The shoe processing support method according to claim 1, further comprising a step of correcting the shape of the processing area corresponding to the welded area in the first shape data based on the shape of the welded area.

5. The shoe processing support method described in claim 1, wherein the step of extracting candidates for the processing area extracts candidates for the processing area based on the outer shape of the joint surface of the first shoe component and the outer shape of at least one of a recess and a protrusion formed on the joint surface of the first shoe component.

6. A shoe processing support method as described in claim 1, wherein the step of extracting the welded area extracts the welded area based on the second shape data pre-stored in a memory unit.

7. The shoe processing support method described in claim 6, wherein the step of extracting the welded area includes a step of correcting the size of the welded area in the second shape data based on the size of the joining surface of the first shoe component and the size of the welded surface of the second shoe component pre-stored in the memory unit.

8. A shoe processing support method as described in claim 1, wherein the step of extracting the welded area extracts the welded area based on the second shape data obtained by imaging the second shoe component.

9. The shoe processing support method according to claim 1, further comprising a step of determining whether the shape of the joining surface of the first shoe component is acceptable or not based on a comparison of the shape of the processing area with the shape of the joined area.

10. The shoe processing support method of claim 9, wherein the step of determining pass / fail determines that the shape of the joining surface of the first shoe component is pass when the deviation between the second moments of area in the foot length direction and foot width direction of the processing area and the second moments of area in the foot length direction and foot width direction of the joined area corresponding to the processing area is below a threshold value.

11. A shoe processing support method as described in claim 1, wherein the step of identifying the processing area identifies the candidate processing area having the smallest deviation between the second moment of area in each of the foot length direction and foot width direction of the candidate processing area and the second moment of area in each of the foot length direction and foot width direction of the joined area as the processing area corresponding to the joined area.

12. A shoe processing support method as described in claim 1, wherein the step of identifying the processing area includes: a first step of calculating a Hu moment for the candidate processing area and a Hu moment for the joined area; and a second step of identifying the candidate processing area having the smallest deviation between the Hu moment calculated for the candidate processing area and the Hu moment calculated for the joined area as the processing area corresponding to the joined area.

13. The shoe processing support method according to claim 12, wherein said first step calculates a Hu moment for said candidate processing area and a Hu moment for said joined area by calculating a plurality of Hu element moments that constitute the Hu moment and have scale invariance, translation invariance and rotation invariance for said candidate processing area and said joined area, and said second step includes a step of calculating a normalized Hu element moment for said candidate processing area and a normalized Hu element moment for said joined area by normalizing the Hu element moment calculated for said candidate processing area and the Hu element moment calculated for said joined area based on a maximum value and a minimum value of the plurality of Hu element moments, and a step of calculating the deviation based on the normalized Hu element moment calculated for said candidate processing area and the normalized Hu element moment calculated for the joined area.

14. The shoe processing support method according to claim 13, wherein the first step calculates the Hu moment for the candidate processing area and the Hu moment for the joined area without using any Hu element moment among the plurality of Hu element moments, which is an invariant that has the property of having its sign inverted with respect to the mirror image shape of the Hu moment.

15. A shoe processing support device comprising: a first acquisition unit that acquires imaging data of a joining surface of a first shoe component; a shape data generation unit that generates first shape data indicating a shape of the joining surface of the first shoe component based on the imaging data; a first extraction unit that extracts candidates for processing areas to be processed for joining to a joining surface of a second shoe component to be joined to the first shoe component by the joining surface based on the first shape data; a second acquisition unit that acquires second shape data indicating a shape of the joining surface of the second shoe component; a second extraction unit that extracts, based on the second shape data, a joining area to be joined to the first shoe component by the joining surface; and an identification unit that identifies the processing areas corresponding to each of the joining areas among the candidates for processing areas based on a comparison between the candidate processing areas and the joining areas.

16. A shoe processing support program that causes a computer to execute the following steps: acquiring image data of a joining surface of a first shoe component; generating first shape data indicating a shape of the joining surface of the first shoe component based on the image data; extracting, based on the first shape data, candidates for processing areas to be processed for joining to a joining surface of a second shoe component to be joined to the first shoe component by the joining surface; acquiring second shape data indicating a shape of the joining surface of the second shoe component; extracting, based on the second shape data, a joining area to be joined to the first shoe component by the joining surface; and identifying, based on a comparison between the candidate processing areas and the joining areas, the processing areas corresponding to each of the joining areas among the candidate processing areas.

Images