Feature point extraction method, control program, and information processing system

Abstract

The feature point extraction method of the web has: a step (c) of receiving input of the stretch ratio of the web generated from the first manufacturing process to the second manufacturing process; a step (d) of comparing the first feature point information of the web in the first inspection data with the second feature point information of the web in the second inspection data using the received stretch ratio; and a step (e) of extracting, based on the comparison result of step (d), a feature point that exists in one of the inspection data and does not exist in the other, or is determined as a defect in one of the inspection data and is not determined as a defect in the other.

Description

Technical Field

[0001] This invention relates to a method for extracting feature points, a control program, and an information processing system. Background Technology

[0002] As liquid crystal displays (LCDs) began to be used in large-screen televisions or large monitors, the required width of the films used on the display surfaces of LCDs also increased. For example, wide films with a width of 2000 mm or more were required. In addition, in order to estimate substrate loss (film loss) in advance or to reduce transportation costs, the manufacture of film rollers with a roll length of 1000 m or more, and further, 3000 m or more, was required.

[0003] In the post-processing stages of manufacturing products based on roll materials such as films, when quality problems such as defects occur, it is necessary to distinguish whether the defect originated in the previous process (i.e., during the manufacturing of the roll material) and was originally present on the roll material, or whether the defect occurred in the post-processing stage. When it is not clear whether the defect occurred in the post-processing stage, improvements to the previous process are sometimes required. To meet these improvement requirements, the specifications of the previous process become more stringent than necessary, resulting in out-of-specification issues. Furthermore, even when the exact impact on the post-processing stage is unknown, sometimes the specifications of the roll material during manufacturing are estimated to be excessively stringent in order to prevent quality problems in advance. This situation also leads to out-of-specification issues. Out-of-specification issues result in decreased yield or increased costs, are environmentally unfriendly, and are detrimental to both the operators of the previous and post-processing stages.

[0004] It is necessary to distinguish whether the defect is inherent in the coil material or occurs in a subsequent processing step. However, this requires ensuring that the coordinate system of the coil material in the previous process corresponds to the coordinate system of the subsequent process. The coil material undergoes extension processes such as uniaxial extension and biaxial extension. In these extension processes, depending on the set conditions, the length of the coil material in the longitudinal direction (conveyor direction) or the short side (width direction) increases by tens to hundreds of percent (hereinafter referred to as "elongation rate").

[0005] When a subsequent process includes such an extension process, the elongation rate needs to be known when aligning the coordinate systems on the roll surface in the preceding and subsequent processes.

[0006] Patent Document 1 describes an inkjet printing apparatus that addresses the issue of reduced print quality caused by ink absorption during printing on substrates such as printing paper and resin films, and by substrate shrinkage during drying. In this apparatus, image data with arranged markers is pre-printed. The positions of these markers and the image data are then used as teacher data to enable a prediction model to learn in advance. By inputting the original document data into this prediction model, a prediction result representing the substrate shrinkage state caused by ink is output before printing.

[0007] Existing technical documents

[0008] Patent documents

[0009] Patent Document 1: Japanese Patent Application Publication No. 2022-110632 Summary of the Invention

[0010] The technology in Patent Document 1 involves printing image data with learning markers arranged on it and using the measured positions of the printed markers to enable a predictive model to learn. A predictive model corresponding to the substrate needs to be prepared.

[0011] The present invention was made in view of the above circumstances, and its object is to efficiently collect information that is helpful for process improvement or shipment specification setting by taking into account the stretching state in both the manufacturing process of the roll material and the manufacturing process of using the roll material for subsequent processing.

[0012] The above-mentioned objectives of the present invention are achieved by the following means.

[0013] (1) A method for extracting feature points of a roll material, comprising:

[0014] Step (a) is to obtain first inspection data in the first manufacturing process of manufacturing roll material or performing post-processing on the manufactured roll material.

[0015] Step (b) is to obtain second inspection data in a second manufacturing process that performs post-processing using the coil material after the first manufacturing process.

[0016] Step (c) is to accept the input of the stretch rate of the roll produced from the first manufacturing process to the second manufacturing process.

[0017] Step (d) involves comparing the first feature point information of the roll material in the first inspection data with the second feature point information of the roll material in the second inspection data using the accepted stretch rate; and

[0018] Based on the comparison results of step (d), step (e) is to extract feature points that exist in both the first and second inspection data, or feature points that exist in one of the first and second inspection data but not in the other inspection data.

[0019] (2) The feature point extraction method described in (1) above, wherein in step (e), feature points present in both the first inspection data and the second inspection data are extracted.

[0020] (3) The feature point extraction method described in (1) above, wherein,

[0021] The first inspection data and the second inspection data are inspection data obtained by processing the images or signals obtained from photographing the roll material to extract feature point information.

[0022] The feature point information includes information about the location of the feature points on the roll material.

[0023] Step (d) includes step (d1) of aligning the position of the roll material in the first inspection data and the second inspection data.

[0024] In step (d1), there is a process of moving the position of a feature point of one of the first inspection data and the second inspection data using the scaling factor.

[0025] (4) The method for extracting feature points described in (1) above, wherein the stretch rate accepted in step (c) is the stretch rate in the width direction, the stretch rate in the length direction and / or the stretch rate in the thickness direction of the roll material.

[0026] (5) The feature point extraction method described in (3) above, wherein,

[0027] The first inspection data and the second inspection data are inspection data obtained by processing the images or signals obtained from photographing the roll material to extract feature point information.

[0028] The feature point information includes information about the size of the feature points and their position on the roll material.

[0029] Step (d) includes step (d1) of aligning the position of the roll material in the first inspection data and the second inspection data.

[0030] In step (d1), there is a process of moving the position of a feature point of one of the first inspection data and the second inspection data in the width direction and / or the length direction using the scaling factor in the width direction and / or the scaling factor in the length direction.

[0031] (6) The feature point extraction method described in (2) above further includes:

[0032] Step (f) calculates the deviation of the feature points extracted in step (e) from each other in the first and second inspection data; and

[0033] Output the calculated deviation step (g).

[0034] (7) The method for extracting feature points described in (1) above, wherein the stretch rate accepted in step (c) is the stretch rate of each section obtained by dividing the width direction of the roll into multiple stages.

[0035] (8) A control program for causing a computer to execute the extraction method described in any one of (1) to (7) above.

[0036] (9) An information processing system, comprising:

[0037] The acquisition unit acquires first inspection data in a first manufacturing process of manufacturing a roll or performing post-processing on the manufactured roll, and second inspection data in a second manufacturing process of performing post-processing on the roll after the first manufacturing process.

[0038] The receiving department receives inputs of the stretch rate of the roll material produced from the first manufacturing process to the second manufacturing process;

[0039] The comparison unit, based on the received stretch rate, compares the first feature point information of the roll material in the first inspection data with the second feature point information of the roll material in the second inspection data; and

[0040] The extraction unit extracts feature points that exist in both the first inspection data and the second inspection data, or feature points that exist in one of the inspection data but not in the other, based on the comparison result of the comparison unit.

[0041] The feature point extraction method of the present invention includes: a step (c) comparing first feature point information of the membrane in first inspection data of a first manufacturing process with second feature point information of the membrane in second inspection data of a second manufacturing process; and a step (d) based on the comparison result of step (c), extracting feature points that exist in one of the first inspection data and the second inspection data but not in the other, or that are determined to be defects in one of the inspection data but not in the other. Thus, information that is helpful for process improvement or shipment specification setting can be efficiently collected both during membrane manufacturing and in manufacturing processes using the membrane for subsequent processing. Attached Figure Description

[0042] The advantages and features provided by one or more embodiments of the present invention can be more fully understood by referring to the following detailed description and accompanying drawings. However, these are for illustrative purposes only and are not intended to limit the invention.

[0043] Figure 1 This is a schematic diagram showing the manufacturing line for the film roller.

[0044] Figure 2 for Figure 1 A top-view diagram of the extended processes, drying processes, and finishing processes of the manufacturing line.

[0045] Figure 3A A schematic diagram showing the structure of the inspection device.

[0046] Figure 3B A schematic diagram showing the structure of the inspection device.

[0047] Figure 3C A schematic diagram showing the structure of the inspection device.

[0048] Figure 4 This is a schematic diagram illustrating an application example of the information processing system of this embodiment.

[0049] Figure 5 This is a table used to illustrate the first to third types of feature points extracted.

[0050] Figure 6 A block diagram illustrating the general structure of an information processing system.

[0051] Figure 7 Examples of various types of data stored in the storage department.

[0052] Figure 8 This is an example of a check data database (DB) stored in the storage department.

[0053] Figure 9This is a flowchart illustrating the process of generating first inspection data during the first manufacturing step.

[0054] Figure 10 This is a flowchart illustrating the process of generating second inspection data during the second manufacturing step.

[0055] Figure 11 This is a flowchart illustrating the feature point extraction process performed in an information processing system.

[0056] Figure 12 This is an example of an operation screen for handling scaling.

[0057] Figure 13 This is a schematic diagram illustrating the feature point extraction process.

[0058] Figure 14 The flowchart of the subroutine for the comparison process in step S35 is shown.

[0059] Figure 15 This is an example of a probability density function representing the location and intensity of a feature point, calculated using kernel density inference.

[0060] Figure 16 This is an example of a list of corresponding points.

[0061] Figure 17 This is an example of a distribution chart showing the degree of deviation.

[0062] Figure 18 This is an example of a distribution chart showing the degree of deviation.

[0063] Figure 19 This is an example of the operation screen in the variation example. Detailed Implementation

[0064] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. However, the scope of the present invention is not limited to the disclosed embodiments. Furthermore, in the description of the drawings, the same reference numerals are used for the same elements, and repeated descriptions are omitted. In addition, the scale of the drawings is exaggerated for ease of explanation and sometimes differs from the actual scale.

[0065] In this embodiment, "roll material" refers to sheet-like materials, including resin films and metal films. Hereinafter, "roll material" refers to a long-length resin film, and the process of processing it will be described as a film roller. The processing also includes processes such as applying coatings and laminating other film-like materials.

[0066] First, refer to Figure 1 , Figure 2 This describes the manufacturing line for film roller 80. Figure 1 This is a schematic diagram showing the manufacturing line for the film roller. Figure 1 , Figure 2The manufacturing line shown illustrates film formation using solution casting, but is not limited to this method; it can also be formed using melt extrusion.

[0067] like Figure 1 As shown, the film roller manufacturing apparatus 1000 includes a casting section 01, a first drying section 02, an extension section 03, a trimming section 05, a second drying section 06, and a winding section 07 (also called a winding device).

[0068] In the film roller manufacturing apparatus 1000, an inspection device 90 is arranged between multiple processes. The inspection device 90 uses the surface of the film F8 being transported on the manufacturing line as a monitoring area to photograph the film surface and analyzes the obtained frame images to detect the transport status of the film F8. The structure of the inspection device 90 will be described later. Figure 3A wait). Figure 1 The hollow arrows in the diagram indicate the direction of transport for membranes, etc.

[0069] The casting section 01 has a mirror-finished strip-shaped metal casting belt (hereinafter referred to as the belt) 01a with an endless support body that travels in a ring (in the direction of the arrow in the figure) and a die 01b on which a casting liquid made by dissolving resin in a solvent is cast onto the belt 01a. In addition, in order to stabilize the casting liquid film flowing out of the die 01b, a pressure reducing chamber (not shown) can be provided on the upstream side of the die 01b relative to the conveying direction of the belt, and a pressure increasing chamber (not shown) can be provided on the downstream side.

[0070] The peeling roller 01d peels off the cast film 01c formed on the tape 01a and generates an unstretched film F8.

[0071] The first drying section 02 (first drying process) includes: a drying box 02a having a drying air inlet 02b and an outlet 02c; and a conveying roller 02d consisting of multiple sets arranged in a vertical group around the conveying film F8.

[0072] In the first drying section 02, the amount of solvent contained in the unstretched film F8 before entering the stretching section 03 (stretching process) can be adjusted. This adjustment of the solvent amount is performed by changing the drying temperature and conveying speed in the first drying section 02 by the control device 08.

[0073] (Extension 03, Trimming 05)

[0074] Figure 2 yes Figure 1 The diagram above shows the extended processes (also known as tenter frame processes), drying processes, and finishing processes of the manufacturing line. Figure 2 , Figure 3AIn this design, the vertical direction is defined as the Z-direction, the conveying direction of membrane F8 is defined as the Y-direction, and the direction orthogonal to the conveying direction and the width direction of membrane F8 is defined as the X-direction. The Y-direction or conveying direction is also called MD (Machine Direction). Furthermore, the X-direction or width direction is also called TD (Transverse Direction) or the left-right direction. Additionally, the Y-direction is also called the length direction, and in relation to the length direction, the width direction is also represented as the width direction.

[0075] The extension section 03 has an MD extension section 03a and a TD extension section 03b. The extension section 03 extends the unextended film F8 conveyed from the first drying section 02. Figure 2 As shown, the extension 03 holds both ends of the heated membrane F8 by a plurality of left and right pairs of holding parts 301. The extension 03 is equipped with holding parts 301, chain, drive part, closure device, opening device, etc. (parts other than holding parts 301 are omitted from the figure).

[0076] In the upstream MD extension section 03a (MD extension process), it accelerates and moves while being conveyed downstream, thereby gradually increasing the moving speed in the conveying direction (in Figure 2 (Indicated by black arrows). Thus, in the MD stretching process, the unstretched membrane F8 is stretched and conveyed along the MD direction.

[0077] In the next stage, TD extension section 03b (TD extension process), the membrane F8 and the like are conveyed by the holding section 301. The holding section 301 is, for example, a clamp, and each of the left and right holding sections 301 is connected to a ring-shaped chain. The holding section 301 moves along the conveying direction by rotating the chain wound on the sprocket by the drive section. A closure device is provided at the upstream end of the TD extension process in extension section 03, and an opening device is provided at the downstream end. The holding sections 301 that are open due to reaching the closure device close sequentially, and those that are open due to reaching the opening device open sequentially. The side end of the membrane F8 is held and conveyed by the holding section 301 in the closed state. In this TD extension section 03b (TD extension process), the left and right pair of holding sections 301 holding the membrane F8 and the like move gradually outward in the width direction while moving along the conveying direction (black arrow). Thus, in the TD stretching process, a pair of holding parts 301 gradually increase the distance (interval) in the width direction, and the membrane F8 and the like held there are stretched and conveyed along the TD direction.

[0078] Furthermore, while an example of sequential extension followed by TD extension is shown in this extension process, it is not limited to this; simultaneous biaxial extension of MD and TD can also be performed. Additionally, tilted extension processes other than TD extension can also be applied. A TD extension process is an extension process that extends in a direction parallel to the width direction of the film. That is, in the TD extension process, the film F8 is extended at an angle of 0° to the width direction. On the other hand, in the tilted extension process, in order to adjust the optical properties of the film, the two ends of the heated film are held by multiple holding members, and it is extended in a tilted direction at an angle within the range of 40 to 50° to the width direction of the film (referred to as "tilted extension"). Furthermore, in this case, the cooling process can include a heat-relieving process for stress relief heat treatment while continuously conveying the extended film.

[0079] The trimming section 05 (trimming process) includes a cutting section 05a and a recycling section 05b. The cutting section 05a comprises two slitting shears 5110L and R on the left and right sides, as well as multiple rollers 5120, etc. The slitting shears 5110L and R are, for example, rotatably supported circular or disc-shaped blades, rotated by a drive unit (not shown). The rotation of the slitting shears 5110L and R can be driven at approximately the same speed in the same direction as the conveying direction of the membrane F8 at the cutting position, or it can be driven in the opposite direction. Through the two slitting shears 5110L and R, the membrane F8 after the cutting position is separated into the central portion of the membrane F8 and the two side ends, namely the trimmed membranes F801 and F802 (also called ears).

[0080] The central portion of the film F8 is the area where the product is produced, and its width ranges from 1000mm to 2500mm. The widths (also called trimming widths) of the trimmed films F801 and F802 are both around tens to hundreds of millimeters (e.g., 170mm). The central portion of the film F8 is conveyed downstream in the conveying direction and supplied to the subsequent winding process. On the other hand, the trimmed films F801 and F802 are conveyed by roller 5120, changing their orientation downward by 90 degrees, and are recovered by the subsequent recovery unit 05b. The recovery unit 05b has a rotating cutter and a suction pipe (neither shown). By means of the rotating cutter 522, the conveyed trimmed films F801 and F802 are cut into small fragments of a few millimeters in size and recovered into the dust collection box in the vacuum cleaner on the downstream side. Furthermore, if the inspection device 90 is configured before the trimming process, the trimming width can also be input to allow for pre-treatment (described later) Figure 11 In step S32), a coordinate transformation corresponding to the trim width is performed based on the photographic data.

[0081] The central portion of the film F8 is conveyed to the downstream winding section 07. A knurling process can be performed between the trimming section 05 and the winding section 07. In the knurling process, knurling is formed at both ends of the trimmed film F8.

[0082] The basic structure of the second drying section 06 (second drying process) is the same as that of the first drying section 02, so the description is omitted.

[0083] (Chapter 07)

[0084] The winding unit 07 includes a winding machine 07a that winds up the knurled film F8 formed in both parts of the trimming unit 05, an air volume control device 07b, a contact or non-contact linear encoder 07c for detecting the travel speed of the extended film F8, a winding shaft speed measuring instrument 07d, a tension control device 07e, and a thickness measuring device 6f.

[0085] like Figure 1 As shown, the casting section 01 dissolves the raw resin in a solvent, and the casting liquid, prepared by adding various additives such as plasticizers, ultraviolet absorbers, anti-deterioration agents, slip agents, and peel accelerators as needed, is ejected from the die head 01b onto the infinitely moving annular belt 01a. Then, the cast film formed on the belt 01a is stripped of solvent on the annular support until a certain degree is reached, and then peeled off from the belt. Next, it is passed through various conveying units to the drying section and the extension section 03 to trim both ends, and after knurling is appropriately formed, it is wound onto the winding shaft at the winding section 07 to manufacture an optical film.

[0086] As Figure 1 , Figure 2 The raw materials used to manufacture the membrane F8 shown are not particularly limited, but are typically polycarbonate resin, polysulfone resin, acrylic resin, polyolefin resin, cycloolefin resin, polyether resin, polyester resin, polyamide resin, polysulfide resin, unsaturated polyester resin, epoxy resin, melamine resin, phenolic resin, diallyl phthalate resin, polyimide resin, polyurethane resin, polyvinyl acetate resin, polyvinyl alcohol resin, styrene resin, cellulose acetate resin, vinyl chloride resin, etc. Furthermore, for example, considering factors such as productivity and quality, the width of the membrane F8 is preferably 1000 mm to 3200 mm. Considering factors such as quality and operability, the thickness is preferably 15 μm to 500 μm.

[0087] Considering productivity, winding quality, etc., the winding is on the winding shaft 82 (refer to...) Figure 1 The length of the optical film F8 on the film roller 80 (also called the winding core) is preferably 2000m to 8000m. The winding length is a value calculated from the speed and time. The winding speed is, for example, 100m / min.

[0088] The inspection device 90 captures images of the film, etc., and generates image data. Figure 1 In the example shown, multiple inspection devices 90 are arranged before and after each step of the extended process of the membrane manufacturing line. Figure 1 In this process, the inspection device 90 is configured to photograph the surface of film F8 immediately after the casting process (casting section 01), immediately after the drying process (first drying section 02), and immediately after the trimming process (trimming section 05). Hereinafter, refer to... Figures 3A to 3C To explain the structure of the inspection device 90.

[0089] (Inspection device 90)

[0090] The following are methods for using transparent materials such as membrane F8 as the object of inspection to detect irregularities on the surface of the material or bubbles, cracks, deformation of the internal structure, etc., inside the transparent material.

[0091] (1) A transmission-type inspection device that detects defects in the inspected object by irradiating it with light and receiving light that passes through the inspected object.

[0092] (2) A reflective inspection device that detects defects in the inspected object by receiving light reflected from the inspected object.

[0093] Furthermore, both the transmissive and reflective types include bright-field inspection devices that receive non-scattered light from the surface based on the positional relationship between the camera's optical axis, the light source, and the inspected object, and dark-field inspection devices that receive scattered light. In a bright-field inspection device, when there is no defect, there is no light scattering, so light from the light source is incident on the light detection unit without obstruction. However, when a defect exists, the light is obstructed and does not incident on the light detection unit. Therefore, the defect is observed as a dark dot or line against a bright background. In contrast, in a dark-field inspection device, when there is no defect, light is not scattered and therefore does not incident on the light detection unit. However, when a defect exists, the light is scattered due to the defect and thus incident on the light detection unit. Therefore, the defect is observed as a bright spot or line against a dark background. The inspection device 90 in this embodiment can be any type of inspection device. Preferably, the first inspection data and the second inspection data are obtained by the same type of inspection device, but this is not a limitation.

[0094] (Reflective inspection device)

[0095] Figure 3A This is a schematic diagram showing the structure of a reflective inspection device 90 as viewed from the width direction (X direction). Figure 3BThis is a schematic diagram showing the structure of the inspection device 90 as viewed from the transport direction (Y direction). The inspection device 90 includes a light source 91, a camera 92 as an optical sensor, an analysis unit 93 as a data processing unit, and a storage unit 94. The inspection device 90 is an apparatus for optically inspecting feature points (hereinafter also referred to as defects) generated in the membrane F8 during transport. In the inspection device 90, the camera 92 optically inspects the membrane F8 of the membrane roller 80 and generates image data as inspection data. The image data includes not only still images but also animation data composed of continuous still images in a time series. Furthermore, as inspection data, feature points can be extracted directly from signal data without image digitization. The number of cameras, the viewing angle, and the distance to the membrane surface are set such that the entire area in the width direction of the membrane F8 becomes the inspection area (shooting range). The number of cameras is arranged so that multiple cameras can be arranged along the width direction when one camera cannot properly capture the entire width of the membrane. Figure 3B This diagram illustrates, as an example, a configuration where two cameras 92 are arranged along the width direction (X direction). The analysis unit 93 can combine multiple images obtained through continuous shooting by a single camera 92 to generate a single image data containing the entire film surface of the film roller 80, or store multiple image data in the storage unit 94 corresponding to the shooting time. Alternatively, it can combine the same image data obtained by multiple cameras 92 arranged along the width direction. The analysis unit 93 can determine the position of the film F8 in the length direction by referring to the stored transport speed (winding speed or delivery speed) and the shooting time corresponding to the image data. Hereinafter, we will describe the case where multiple image data obtained through continuous shooting corresponding to a single film roller 80 are stored corresponding to the shooting time. The analysis unit 93 generates defect information by analyzing the image data. The inspection device 90 inspects defects generated during manufacturing processes such as the winding process of the long-sized film F8.

[0096] Light source 91 illuminates the inspection area of ​​membrane F8. Light source 91 is a light source that illuminates the roll-shaped membrane F8 uniformly in the width direction (orthogonal to the length direction of membrane F8 and parallel to the membrane surface). Here, uniform means that the illuminance in membrane F8 is approximately the same in the width direction of membrane F8 (the difference between the maximum and minimum values ​​is less than a predetermined value, etc.).

[0097] Camera 92 is an optical sensor that optically reads the inspection area of ​​film F8. Camera 92 includes an imaging element such as a CCD (charge-coupled device) or CMOS (complementary metal-oxide-semiconductor), a lens, etc. Camera 92 is an area sensor that generates two-dimensional image data based on the output signals of each imaging element. Camera 92 detects diffuse light in the light illuminated by light source 91 and reflected in the inspection area of ​​film F8. Here, either a color camera or a black-and-white camera (monochrome camera) can be used as camera 92.

[0098] One or more cameras 92 have a shooting range covering the entire width direction of the film F8, and simultaneously read the entire width direction of the film F8 in a single shot. The camera 92 can be a camera that detects light in the visible light region or a camera that detects light in the infrared region.

[0099] Furthermore, it is preferable that, in the output signal of the camera 92, the contrast between the signal value on the film F8 corresponding to the illuminated portion illuminated by the light source 91 and the signal value corresponding to the non-illuminated portion not illuminated by the light source 91 is a predetermined value or higher. That is, it is preferable that only the areas on the film F8 illuminated by light from the light source 91 (illuminated portions) are brightly visible.

[0100] Contrast ratio is expressed as the difference or ratio of two values ​​of the processed object (here, the signal value corresponding to the irradiated part and the signal value corresponding to the non-irradiated part). The greater the difference between the two values, the greater the contrast ratio. In order to increase the contrast ratio between the irradiated part and the non-irradiated part, a powerful light source 91 with high linear propagation is preferably used.

[0101] Here, "powerful" means that when the illuminance at a distance of 50mm is E50, the illuminance E50 is above 50000lx. "High linear propulsion" means that when the illuminance at a distance of 50mm is E50 and the illuminance at a distance of 100mm is E100, (E50-E100) / E50<0.5 is satisfied.

[0102] The analysis unit 93 consists of a CPU, RAM, etc., and reads various processing programs stored in the storage unit 94 and expands them into RAM, and performs various processes in cooperation with the program.

[0103] The storage unit 94 is composed of HDD, SSD (solid-state drive), etc., and stores various processing programs and data required for their execution. Furthermore, the storage unit 94 stores the captured image data (inspection data) in association with the capture time. The storage unit 94 stores the winding speed (e.g., 100 m / min) in the film roller manufacturing apparatus 1000, or the film F8 pull-out conditions (e.g., 30 m / min) in the product manufacturing apparatus 2000. These winding speeds and pull-out conditions can also be included in the inspection list of the inspection DB (see reference). Figure 8 Table T3).

[0104] The analysis unit 93 detects feature points (position / intensity) such as defects in the film F8 by performing data processing on the output signal of the camera 92 (optical sensor). The data processing includes: image processing of image data obtained from the output signal of the camera 92; defect determination processing to determine defects based on the image-processed data; and quantitative evaluation processing to quantitatively evaluate defects based on the image-processed data.

[0105] (Relative position of camera 92 and light source 91)

[0106] The camera 92 can also be positioned to receive orthogonally reflected light from the light source 91 (in the case of a bright-field inspection mode that receives non-scattered light).

[0107] Furthermore, the camera 92 can also be positioned to avoid receiving reflected light from the light source 91 (in the case of a dark-field inspection mode that receives diffused light). That is, it is preferable to position it to receive diffused light reflected from the object being inspected.

[0108] (Transmission-type inspection device)

[0109] Figure 3C This is an example of a transmissive inspection device 90. A transmissive inspection device 90 can also be used in which the light source 91 is positioned opposite the camera 92 via the membrane F8.

[0110] Figure 4 This is a schematic diagram illustrating an application example of the information processing system 50 of this embodiment. For example... Figure 4 As shown, the information processing system 50 is interconnected with terminal devices 70 in factories A and B via a network. The network is a communication line such as a data communication network. In some networks, wired LANs or wireless LANs (e.g., LANs conforming to the IEEE 802.11 standard) may also be used.

[0111] Terminal device 70 is, for example, a PC (personal computer). For example, terminal device 70 is a PC used by employees of the manufacturing company operating factory A and factory B.

[0112] Factory A is equipped with the aforementioned film roller manufacturing apparatus 1000. Factory A is operated or managed, for example, by a film manufacturer. In factory A, the first manufacturing process of manufacturing film roller 80 is carried out. The film surface of film roller 80 is inspected by inspection device 90.

[0113] Factory B is equipped with a product manufacturing facility 2000. Factory B is operated or managed by, for example, a coating manufacturer (hereinafter also referred to as a user company or user). There are multiple user companies operating each factory B. In factory B, products are manufactured using film rollers 80 shipped and transported from factory A. In factory B, a film (hereinafter, film F8) is pulled out from the film rollers 80, and a first manufacturing step or a second manufacturing step is carried out as a post-processing step, such as coating, or laminating or bonding with other films.

[0114] In this embodiment, as a typical example, the process of manufacturing the film (roll) is referred to as the first manufacturing process, and the post-processing performed on the manufactured film at factory B is referred to as the second manufacturing process. For example, in the post-processing process, a stretching process or a stretching process and a coating process that imparts a functional layer to the surface are performed. In the second manufacturing process, the film surface of the film roller 80 is also inspected by the inspection device 90.

[0115] As an example, except for typical exceptions, when the manufactured film undergoes multiple post-processing steps in factory B, the earlier post-processing step is sometimes referred to as the first manufacturing step, and the subsequent post-processing step is sometimes referred to as the second manufacturing step. Furthermore, sometimes the film roller 80 manufactured in factory A is pulled out and post-processed in a later process. In this case, the process of manufacturing the film roller 80 is sometimes referred to as the first manufacturing step, and the post-processing performed after pulling out the film roller 80 within the same factory A is sometimes referred to as the second manufacturing step. The second manufacturing step performed within factory A includes stretching and drying processes.

[0116] (Overview of feature point extraction processing)

[0117] Although details regarding feature point extraction will be described later, the following references... Figure 4 This section provides an overview of the feature point extraction process. As a typical example, the process of manufacturing the film (roll material) will be referred to as the first manufacturing process, and the post-processing performed on the manufactured film at factory B will be referred to as the second manufacturing process.

[0118] In the first manufacturing process of factory A, during product inspection, the film is optically inspected by inspection device 90 to generate inspection data (also known as fault data or defect data). Specifically, by analyzing image data obtained from photographing the surface of the film, inspection data containing feature point information (feature point, position, intensity) is generated (hereinafter referred to as first inspection data). Information processing system 50 obtains the first inspection data from terminal device 70 of factory A (step S1).

[0119] Here, feature points refer to defects on the membrane, generated by analyzing image data. Image analysis can utilize known techniques to extract pixels whose pixel values ​​deviate from the surrounding average by a predetermined margin (difference exceeding a predetermined margin) from image data of the membrane surface, or it can be calculated using the "image processing for feature point generation" method described below. In most cases, one or more image data obtained from photographing a membrane roller 80 (several km long) generate tens to thousands of feature points. Defects include both problems at the level of product defects and minor problems at the level of non-product defects. Feature points include defects related to poor adhesion, axial misalignment, etc., during the bonding of membranes together (e.g., ultrasonic welding). Feature point information includes size and position (xy coordinates). Furthermore, information that aggregates (clusters) multiple adjacent feature points into one can also be used as feature point information. The image processing for feature point generation will be described later.

[0120] The film roller 80 manufactured in factory A is transported to factory B. In the second manufacturing process of factory B, during product inspection, the film on the film roller 80 is optically inspected by inspection device 90, generating inspection data. By analyzing image data obtained from photographing the surface of the film, inspection data containing feature points (hereinafter referred to as second inspection data) is generated. Information processing system 50 obtains the second inspection data from terminal device 70 in factory B (step S2). The inspection device 90 in factory A (first manufacturing process) and the inspection device 90 in factory B (second manufacturing process) are preferably the same, that is, they have the same measurement system and the same measurement conditions, but are not limited thereto. In factory A and factory B, the required performance, quality, product specifications (hereinafter referred to as product specifications, etc.) may sometimes be different, and an inspection device 90 with an appropriate measurement system and measurement conditions may be used according to the respective product specifications, etc.

[0121] The information processing system 50 performs feature point extraction processing (step S3) by comparing first and second inspection data related to the same film roller 80. Specifically, the information processing system 50 performs feature point extraction processing by comparing feature points at corresponding positions on the film surface in the first and second inspection data. In this specification, detecting feature points from image data is referred to as "generating feature points". Classifying feature points into any of the following first to third types of feature points by comparing the first and second inspection data is referred to as "extracting feature points".

[0122] Figure 5This table illustrates the first to third types of feature points extracted through the feature point extraction process. Multiple feature points generated from inspection data of a film roller 80 are sometimes classified into the first to third types of feature points. For example, some of several hundred feature points are extracted as the third type, thereby extracting the first and second type feature points.

[0123] (First type of feature point)

[0124] The first type of feature point is a feature point that exists in the first inspection data but not in the second inspection data. This first type of feature point is a feature point that disappears in the second manufacturing process (e.g., the coating process). This first type of feature point is a feature point that does not require management in the first manufacturing process. In this case, the manufacturing conditions that cause the generation of the first type of feature point can be subject to regulatory leniency in the first manufacturing process.

[0125] (Second type of feature point)

[0126] The second type of feature point is a feature point that is not present in the first inspection data but exists in the second inspection data. This second type of feature point is a newly generated feature point in the second manufacturing process. Since this second type of feature point originates from the second manufacturing process, it can be used to improve the second manufacturing process.

[0127] (Third type of feature point)

[0128] The third type of feature point exists in both the first and second inspection data. This third type of feature point originates from the first manufacturing process and is a feature point that needs to be managed. Because this third type of feature point originates from the first manufacturing process, it can be used to improve the first manufacturing process.

[0129] The information processing system 50 can also be configured to provide feedback on the feature point extraction results to users such as staff at factory A and factory B (step S4). For example, in response to access from the terminal device 70 of the user of the object's film roller 80 and the terminal device 70 of the user to which the film roller 80 is delivered, the extraction results are sent to the terminal device 70. The above is an overview of the feature point extraction process. More details about the process are described below.

[0130] (Information Processing System 50)

[0131] The following is for reference. Figures 6-8 Description of Information Processing System 50. Figure 6 This is a block diagram illustrating the general structure of the information processing system 50. The information processing system 50 is, for example, a server. Figure 6 As shown, the information processing system 50 includes a control unit 51, a storage unit 52, and a communication unit 53.

[0132] (Control Department 51)

[0133] The control unit 51 includes a CPU and memory such as RAM and ROM. The CPU is a control circuit composed of a multi-core processor that executes the control and various arithmetic operations of the above-mentioned units according to the program. The various functions of the information processing system 50 are performed by the CPU executing its corresponding program.

[0134] The control unit 51 functions as an acquisition unit 511 and a receiving unit 512 in cooperation with the communication unit 53. Furthermore, the control unit 51 functions as a comparison unit 513, an analysis unit 514, an extraction unit 515, and an output unit 516. The acquisition unit 511 acquires first inspection data and second inspection data obtained through inspections in the first and second manufacturing processes. The receiving unit 512 receives input of the scaling factor from the user. The comparison unit 513 extracts feature points from each of the first and second inspection data and performs comparison processing using the received scaling factor. The comparison unit 513 searches for corresponding feature points between the first and second inspection data through comparison processing. In the comparison processing, the comparison unit 513 generates descriptors for the feature points (descriptor 2 described later) for the two inspection data (image data), uses these descriptors to perform feature point matching processing between the inspection data, and outputs the comparison result (described later). Figure 16 (See the corresponding point list shown). The analysis unit 514 uses the corresponding point list to calculate the deviation index. The deviation index includes statistical information such as scatter plots and correlation coefficients. The extraction unit 515 uses the comparison results to extract (classify) the first to third type of feature points. In response to a request from the terminal device 70, the output unit 516 sends the feature point extraction results to the terminal device 70 or displays them on a display unit (not shown).

[0135] (Storage Department 52)

[0136] Storage unit 52 is a large-capacity auxiliary storage device that stores various programs and data, including the operating system. The storage device may be, for example, a hard disk drive, a solid-state drive, flash memory, or ROM. Storage unit 52 stores user lists, batch lists, and inspection data databases. The management and registration of user lists and batch lists are performed through access to an administrator's terminal device 70. For example, this administrator might be the head of the department responsible for operating factory A.

[0137] (User List)

[0138] Figure 7 These are examples of various data stored in storage unit 52. Figure 7Table T1 shown is an example of a user list. The user list stores user IDs, usernames, contact information, etc. Additionally, users are given access to the search data DB (check database) for each user, and are granted access to various data (check data, extract data, etc.) related to the membrane roller 80 (identified by batch ID) associated with the user.

[0139] (Batch list)

[0140] Figure 7 Table T2 is an example of a batch list. The batch list records the batch ID assigned to each film roller, the product name (also known as variety), the delivery destination user ID (the ordering party), and multiple manufacturing conditions, dimensions (width, length, thickness), manufacturing date, etc.

[0141] (Check the database)

[0142] Figure 8 This is an example of a check database stored in storage unit 52. The check database stores... Figure 8 The data shown includes various inspection-related data for the film roller 80, such as the first inspection data, the second inspection data, and the feature point extraction results. As mentioned above, the first inspection data is data obtained during inspection in the first manufacturing process. The second inspection data is data obtained during inspection in the second manufacturing process. The feature point extraction results are data generated by the information processing system 50 using these first and second inspection data.

[0143] Figure 8 Table T3 shown is an example of an inspection list registered in the inspection data DB. The inspection list stores information such as inspection ID, batch ID, inspection device ID, inspection data, and inspection date and time.

[0144] Figure 8 Table T4 shows an example of the contents of the inspection data (inspection ID: i0101) in the inspection list. The inspection data records the feature point IDs that are automatically assigned to each feature point in sequence, as well as feature point descriptors 1, 2 (hereinafter referred to as descriptor 1, etc.) for each feature point ID.

[0145] Descriptor 1 contains individual information about the feature point, recording its XY coordinate position and intensity. Intensity refers to the feature point's level, which will be described later. Additionally, as intensity information, the size (diameter, area) of the feature point may also be included. The XY coordinate position is based on the starting point of the membrane surface (e.g., the left end of the front end). X is the coordinate in the width direction of the membrane, and can take a range of, for example, 0 to 3000 mm, corresponding to the membrane size (see Table T2). Y is the coordinate in the length direction of the membrane, and can take a range of, for example, 0 to 10000 m, corresponding to the membrane size. The generation of descriptor 1 is performed in the analysis unit 93 of the inspection apparatus 90.

[0146] Descriptor 2 is surrounding information, which is a vector or arrangement of information representing the relationship with other feature points. For example, SIFT feature values ​​can be used as descriptors, or the probability density function of feature points calculated by performing kernel density inference can be used as a descriptor. The generation of descriptor 2 is mainly performed by the comparison unit 513.

[0147] Figure 8 Table T5 is an example of the extracted result data (hereinafter referred to as the extracted data). The extracted data records the inspection IDs of the first and second inspection data (which served as the original data) and the extraction result for each feature point. The extracted results (first to third types) are as described above. Figure 5 The classification is shown below. The composite feature point ID is automatically assigned sequentially, generating composite feature points corresponding to feature points present in either or both of the first and second inspection data. The number of composite feature point IDs is greater than or equal to the number of feature point IDs in the first and second inspections.

[0148] Furthermore, for a single batch ID, multiple first inspection data and second inspection data generated by multiple inspection devices may sometimes be generated. For example, there may be situations where, during a second manufacturing process, while inspecting (photographing) the rolled-up film roller 80, inspections are also performed in some downstream processes, thereby generating multiple second inspection data. In this case, the information processing system 50 can also be configured to generate multiple feature point extraction result data (one-to-many) between multiple second inspection data for a single first inspection data. Additionally, the user can select which second inspection data corresponds to (e.g., as described later). Figure 12 Buttons b6, b7, etc.

[0149] (Communications Department 53)

[0150] The communications section 53 is also an interface for network connection with external devices such as PCs.

[0151] (The generation and processing of the first and second inspection data)

[0152] The following is for reference Figure 9 , Figure 10 The generation and processing of the first and second inspection data performed in the first and second manufacturing processes are explained. Figure 9 This is a flowchart representing the generation process of the first inspection data in the first manufacturing process.

[0153] (Generation and processing of the first inspection data)

[0154] (Step S11)

[0155] In the above-described typical example, in the first manufacturing process, the film roller 80 is manufactured by the film roller manufacturing device 1000.

[0156] (Step S12)

[0157] The inspection device 90 captures images of the film and saves the image data. The inspection device 90 is... Figure 3A As explained in the document.

[0158] (Step S13)

[0159] The analysis unit 93 performs the image processing described below on the image data and generates multiple feature points.

[0160] (Image processing for feature point generation)

[0161] The analysis unit 93 acquires two-dimensional image data generated by the camera 92 and stored in the storage unit 94.

[0162] The analysis unit 93 performs data processing on the image data (inspection data) acquired from the camera 92.

[0163] The analysis unit 93 divides the image data into multiple regions. For example, the analysis unit 93 divides the image data into n (e.g., several to dozens) regions (hereinafter referred to as region a1 to region an) in the width direction.

[0164] Next, the analysis unit 93 acquires image data for region a1 and performs mathematical processing on the image data for region a1. Appropriate mathematical processing is prepared based on the type of defect in the detected object (thickness band, longitudinal wrinkles, diagonal wrinkles, etc.).

[0165] Mathematical processing includes preprocessing, emphasis processing, signal processing, and image feature extraction.

[0166] As a preprocessing step, the following are listed.

[0167] Image cropping

[0168] Low-pass filter, high-pass filter, Gaussian filter, median filter, bilateral filter.

[0169] • Morphological transformations, color transformations (L*a*b*, sRGB, HSV, HSL), contrast adjustment, noise removal, restoration of blurred / shaky images, mask processing, Hough transform, projective transformation, etc.

[0170] As an example of emphasizing the processing, examples include the Sobel filter, Scharr filter, Laplacian filter, Gabor filter, and Canny method.

[0171] As a signal processing technique, the following are examples.

[0172] • Calculate basic statistics (maximum, minimum, mean, median, standard deviation, variance, quartiles), sum of squares and square roots, differences, sums, products, ratios, distance matrix processing, differentiation and integration, threshold processing (binarization, adaptive binarization, etc.).

[0173] • Fourier transform, wavelet transform, peak detection (peak value, number of peak values, half-width, etc.), etc.

[0174] Examples of image feature extraction methods include template matching and SIFT features.

[0175] Next, the analysis unit 93 performs thresholding on the values ​​(feature quantities) obtained through mathematical processing of the image data for region a1. Thresholding is a process that determines whether a defect is a target for detection and determines the level (intensity) of the defect based on a predetermined threshold.

[0176] In threshold processing, determining the existence and type of defects corresponds to "defect determination processing." Additionally, in threshold processing, classifying defects into multiple levels based on thresholds corresponds to "quantitative evaluation processing."

[0177] For example, for parameters (features) with values ​​from 1 to 100, defects are classified into multiple levels. For instance, levels can be classified based on defect size (diameter or area). Furthermore, the size-based classification can be further subdivided based on the parameter value.

[0178] Analysis unit 93 performs the same processing on areas other than area a1.

[0179] After processing each region a1 to an, the analysis unit 93 integrates the results for each region a1 to an and ends the data processing. Specifically, the analysis unit 93 generates data for each region (each position in the width direction of the membrane F8) that corresponds to the level of the detected defect and its location (xy coordinates).

[0180] After data processing, the analysis unit 93 saves the processing results to the storage unit 94. The analysis unit 93 performs data processing on each of the multiple image data obtained from inspection by a film roller 80 to obtain a processing result. By summarizing these processing results, a result is generated as follows: Figure 8 The inspection data is shown in Table T4.

[0181] (Step S14)

[0182] The terminal device 70, present in the first manufacturing process, sends inspection data containing information on multiple feature points obtained through the processing up to step S13 to the information processing system 50. The acquisition unit 511 of the information processing system 50 stores the acquired inspection data as first inspection data in the inspection data DB of the storage unit 52.

[0183] (Generation and processing of the second inspection data)

[0184] Figure 10 This is a flowchart illustrating the generation process of the second inspection data performed in the second manufacturing process.

[0185] (Step S21)

[0186] In the second manufacturing process, after the first manufacturing process, the product manufacturing apparatus 2000 performs post-processing using the film roller 80 to manufacture a product using the film F8.

[0187] (Step S22)

[0188] The inspection device 90 takes pictures of the surface of the membrane F8 before post-processing, during post-processing, or after post-processing, and saves the image data.

[0189] (Step S23)

[0190] The analysis unit 93 performs the same process as in step S13, storing the inspection data, including feature point information related to the generated multiple feature points, into the storage unit 94.

[0191] (Step S24)

[0192] In the second manufacturing process, the terminal device 70 sends inspection data, including information on multiple feature points obtained through the processing up to step S23, to the information processing system 50. The acquisition unit 511 of the information processing system 50 saves the acquired inspection data as second inspection data to the inspection data DB of the storage unit 52. The second inspection data is similarly described by feature point id and feature point descriptors 1 and 2, as shown in Table T4.

[0193] (Feature point extraction and processing)

[0194] The following is for reference. Figures 11-18 This describes the feature point extraction process performed by the information processing system 50. Figure 11 This is a flowchart representing the feature point extraction process. Figure 12 This is an example of an operation screen displayed on the terminal device 70. Figure 13 This is a schematic diagram illustrating the feature point extraction process.

[0195] The information processing system 50 responds to a start instruction from the user via the operation screen of the terminal device 70, or to the timing when the second check data is registered in the check data DB of the storage unit 52 and a pair of the first check data and the second check data are complete, and begins the processing of step S31 and below. Figure 12 This is an example of an operation screen 701 displayed on the terminal device 70. After selecting a batch, the user selects first inspection data and second inspection data from multiple inspection data associated with that batch via buttons b6 and b7. The second inspection data is inspection data obtained in a process downstream of the first inspection data. Furthermore, in this embodiment, it is described as if there is an extension process between the locations where the first and second inspection data are collected.

[0196] (Step S31)

[0197] The acquisition unit 511 acquires the same batch, i.e., a pair of first and second inspection data, from the inspection data DB. The same batch is associated based on the batch ID assigned to the inspection ID. Furthermore, the batch ID can be assigned a branch number ID or a new batch ID in any process; it may not be a one-to-one correspondence, but sometimes a pair of n or n-to-one. For example, sometimes a 3000m film from a previous process is divided into three parts of 1000m in a subsequent process. In this case, the branch number or a new batch ID is assigned to the batch ID in the subsequent process. Conversely, sometimes two or three film rollers are joined into one film in a previous process and used in a subsequent process. In this case, there are two or three batch IDs in the previous process and one batch ID in the subsequent process. In either case, the corresponding association between the batch IDs of each process is described in the batch list (see reference). Figure 7 )middle.

[0198] (Steps S32 and S33)

[0199] The comparison unit 513 performs preprocessing on the first inspection data under the first condition and preprocessing on the second inspection data under the second condition.

[0200] Comparison section 513, etc. Figure 13As shown, as a preprocessing step for the second condition, if there is a difference between the winding (first manufacturing process) and pulling (second manufacturing process) of the second inspection data, a preprocessing step of reversing the Y-coordinate (vertical) is performed to align it. Furthermore, in the second manufacturing process, based on information about which side of the film F8, the surface or the back side, is designated as the shooting area for the camera 92, the comparison unit 513 performs a preprocessing step of reversing the X-coordinate (left-right) for the second inspection data (or the first inspection data).

[0201] In addition, the comparison unit 513 performs at least one of the following as noise removal processing included in the first and second conditions for the first inspection data and the second inspection data.

[0202] (1) Remove low-level feature points.

[0203] (2) Remove extremely small feature points.

[0204] (3) Remove continuous dots.

[0205] (4) Remove concentrated dots in the width direction. This is especially common at the front or back of membrane F8.

[0206] (Step S34)

[0207] The acceptance department 512 adjusts the user acceptance rate. For example, this is displayed on the screen... Figure 12 The operation screen 701 on the terminal device 70 shown accepts input of the scaling ratio. The user can input any value within a predetermined range (e.g., 50-300%) via the keypad using buttons b1 and b2. The receiving unit 512 accepts input of the scaling ratio in the Y direction (length direction) via user input to button b1. Additionally, the receiving unit 512 accepts input of the scaling ratio in the X direction (width direction) via user input to button b2. Figure 12 The operation screen 701 shows states where 140% has been entered. When the value is greater than 100%, it is assumed that the membrane in the second inspection data is longer than the membrane in the first inspection data. The user sets the stretch rate based on the post-processing (second manufacturing step) condition settings. These condition settings are described in Table T2, which is associated with the batch list, and can be referenced by the user. For example, if the stretch rate can be predicted in advance when the membrane F8 stretches based on the heating temperature or stretching settings in the post-processing of the second manufacturing step, the user enters this stretch rate.

[0208] Furthermore, extension 03 performs TD extension in the X direction and MD extension in the Y direction, therefore in Figure 12Examples of receiving elongation rates in the X and Y directions are shown. However, this is not a limitation. For example, when a slant stretching process is performed in the stretching section 03, the receiving section 512 can also receive the elongation ratio (elongation rate) of the slant stretching. For example, when the film F8 is used as a phase difference film, a slant stretching process is performed. In this case, the receiving section 512 receives an input of an stretching angle value within a predetermined range (40 to 50° or 130 to 140°). Alternatively, instead of this stretching angle, the receiving section 512 can also receive the degree of deformation caused by shearing. For example, it receives the elongation ratio in the orientation angle and orientation angle direction of the roll material, calculated using the bending angle during stretching and the MD and TD direction elongation ratios.

[0209] (Step S35)

[0210] The comparison unit 513 performs comparison processing to search for feature points in the other party's inspection data that are the same as or correspond to the feature points in the inspection data of one party, and performs correspondence (matching) between the feature points. Figure 14 This is a flowchart illustrating the subroutine processing of step S35.

[0211] (Step S401)

[0212] The comparison unit 513 moves the position of a feature point in one of the first and second inspection data according to the scaling factor received in step S34. That is, the comparison unit 513 transforms the coordinate system. Hereinafter, the explanation will assume that the comparison unit 513 moves the feature point of the first inspection data according to the scaling factor. For example, in... Figure 12 In the input example, the comparison unit 513 converts the Y-coordinate of each feature point of the first inspection data into a value multiplied by the stretch rate based on the origin (0). Here, the origin is the leading position (the starting position of the roll-up) of the film F8 during the production of the film roller 80 (the last position when unrolled). Furthermore, the comparison unit 513 converts the X-coordinate of each feature point of the first inspection data into a value multiplied by the stretch rate based on the center point. That is, the first inspection data is converted to extend to both sides based on the center point. Here, the center point is the center position of the film in the width direction of the first inspection data. As another example, in the case of oblique extension, based on the input extension angle and elongation rate, the comparison unit 513 calculates the elongation rate (stretch rate) in the X direction and the elongation rate (stretch rate) in the Y direction corresponding to the position in the width direction of the film in the first inspection data, and applies this. Additionally, as another example, when the orientation angle and elongation rate in the orientation angle direction of the roll material are received, the elongation rates in the X and Y directions calculated therefrom are calculated and applied. Furthermore, as another example, when transforming the second inspection data side, in addition to dividing by the scaling factor, the comparison unit 513 can perform coordinate transformation by the same process as when transforming the first inspection data side.

[0213] (S402)

[0214] The comparison unit 513 generates a descriptor 2 for each feature point of the first and second inspection data. The comparison unit 513 may use SIFT feature values ​​as descriptors, or it may use the probability density function of the feature points calculated by performing kernel density inference as a descriptor.

[0215] (Steps S403, S404)

[0216] The comparison unit 513 compares the feature points of the first and second inspection data with each other and searches for the most corresponding point. The similarity between the feature points is evaluated by descriptor 1 and descriptor 2, and the most similar point is regarded as the corresponding feature point.

[0217] For example, in comparing feature points of the first and second inspection data, the comparison unit 513 searches for feature points in the first inspection data that correspond to the object feature points of the second inspection data. At this time, the comparison unit 513 determines the feature points of the first inspection data that have the same X and Y coordinates as or have the closest intensity to the feature points of the feature points in the second inspection data as the same point. Alternatively, it determines the feature points whose X and Y coordinates are closest in distance (Euclidean distance) as the same point (corresponding points). The comparison unit 513 excludes feature points whose distance is farther than a set threshold from the determination of corresponding points. Furthermore, in this embodiment, regarding the threshold, the threshold in the Y direction is set to a value 2 to 3 decimal places larger than that in the X direction. For example, the threshold in the X direction is a few millimeters, and the threshold in the Y direction is a few meters.

[0218] Furthermore, the comparison unit 513 can use descriptor 1 and descriptor 2 to extract the most similar feature points from each other by measuring the distance between vectors in the high-dimensional vector space. In this case, descriptor 2 can use a probability density function calculated by kernel density inference as described above. Figure 15 This is an example of a probability density function representing the location and intensity (density) of feature points, calculated through kernel density inference. Figure 15 In the diagram, the horizontal and vertical axes represent the X and Y coordinates, and the higher the concentration, the higher the density.

[0219] The comparison unit 513 determines that each feature point corresponds to only one feature point. For example, for feature points in the second inspection data, the feature point whose descriptor vectors in the first inspection data are closest to each other is registered in the corresponding point list.

[0220] Figure 16This is an example of a list of corresponding points stored in storage unit 52. The list of corresponding points is a list that maps the most similar feature points in the first inspection data to each feature point in the second inspection data. Furthermore, the list of corresponding points describes the XY coordinates of the feature points in the second inspection data, as well as the Euclidean distance between the corresponding feature points, the difference dx in the X coordinate, and the difference dy in the Y coordinate.

[0221] (Step S405)

[0222] The analysis unit 514 generates the deviation between the feature points existing in both the first inspection data and the second inspection data. That is, for the feature points in the corresponding point list, the analysis unit 514 generates any one of the following four scatter plots as data representing the deviation.

[0223] (1) The x-axis is the coordinate of the x-axis, and the y-axis is the dx-axis.

[0224] (2) Horizontal axis X coordinate, vertical axis dy,

[0225] (3) Horizontal axis Y-coordinate, vertical axis dy,

[0226] (4) Horizontal axis Y coordinate, vertical axis dx.

[0227] Figure 17 Here is an example of a scatter plot corresponding to (3) above. Figure 18 This is an example of a scatter plot corresponding to (1) above. The horizontal axis shows the Y-coordinate (or X-coordinate) of the feature points in the second inspection data. Furthermore, the analysis unit 514, as a deviation, can replace the scatter plot or be used in conjunction with the scatter plot. Figure 1 It is used to calculate statistical information (e.g., regression line, correlation coefficient, etc.). Figure 18 and Figure 19 The regression line for the scatter data is also shown in the diagram.

[0228] (Step S36)

[0229] The output unit 516 outputs the deviation. For example, a scatter plot of the deviation generated in step S405 is displayed. Figure 12 The preview area b5 of the operation screen 701 is shown. Additionally, correlation coefficients can be overlaid and displayed in the preview area b5. Users can refer to... Figure 12 The deviation shown in preview area b5 is used to identify whether the entered scaling factor is appropriate.

[0230] (Step S37)

[0231] When the user identifies that the scaling ratio needs to be corrected (Yes), the step involves inputting the corrected scaling ratio in the operation screen 701. In response to receiving the new scaling ratio in step S34, the control unit 51 repeats the processing after step S34.

[0232] On the other hand, if the user determines that no adjustment to the scaling ratio is needed (No), the user presses the registration button b4 on the operation screen 701. The control unit 51 responds to the operation of the registration button b4 and proceeds to step S38.

[0233] (Step S38)

[0234] The extraction unit 515 classifies feature points according to the correspondence list, the first inspection data, and the second inspection data. Specifically, the extraction unit 515 classifies the corresponding feature points in the correspondence list as third type feature points. In addition, the extraction unit 515 classifies feature points in the second inspection data that do not correspond to feature points in the first inspection data as second type feature points in the correspondence list.

[0235] Furthermore, the extraction unit 515 classifies feature points in the first inspection data that are not included in the corresponding list (feature points that do not correspond to feature points in the second inspection data) as first type feature points (see [link]). Figure 5 and Figure 8 Table T5).

[0236] (Step S39)

[0237] The output unit 516 registers the extraction results (extracted data) generated in step S38 into the inspection data DB, or sends the extraction results to the terminal device 70. End of section. Figure 11 The feature point extraction process shown is now complete.

[0238] Thus, in this embodiment, the input of the stretch rate of the roll material is used to compare the first feature point information of the film in the first inspection data with the second feature point information of the film in the second inspection data. Based on this comparison result, feature points that exist in one inspection data but not in the other, or that are identified as defects in one inspection data but not in the other, are extracted from the first and second inspection data. Therefore, information that contributes to process improvement and shipment specification setting can be efficiently collected during both the manufacturing of the film roll and the subsequent manufacturing processes using this information. Especially when there is an extension process between the first and second manufacturing processes, feature point comparison can be correctly performed using the input stretch rate, thereby enabling accurate feature point extraction.

[0239] (Variation Example 1)

[0240] In the above embodiment, an example is shown where a value is set in each of the X and Y directions as the elongation rate. However, it is also possible to divide it into multiple stages as shown below and accept the elongation rate of each zone.

[0241] Figure 19 This diagram shows the operation screen 702 displayed on the terminal device 70 in the modified example. On the operation screen 702, the user can divide the X direction (width direction) into five equal sections using button b20 and input the stretch rate individually for each section. Similarly, in the Y direction (length direction), the user can divide it into five equal sections using button b10 and input the stretch rate individually for each section. In the stretching process, the film is stretched while being heated; however, depending on the process structure, uneven heating and temperature distribution can cause deviations in the stretch rate depending on the position of the roll surface. For example, sometimes the temperature at the center and ends in the width direction differs. In the modified example, in step S34, the receiving unit 512 receives the input of a fine stretch rate based on the position in the width direction. Then, in step S401, the comparison unit 513 divides the X coordinates of each feature point of the first inspection data into sections based on the center point and transforms them into values ​​multiplied by the stretch rate set for each section. The receiving unit 512 and the comparison unit 513 also perform the same processing in the length direction.

[0242] Thus, in the modified example, by accepting the scaling factor corresponding to the division in the width direction, the comparison of feature points can be performed more accurately and correctly, thereby enabling the extraction of feature points more accurately.

[0243] The structure of the information processing system 50 described above is based on the main structure described in the description of the features of the above embodiments. It is not limited to the structure described above, and various modifications can be made within the scope of the claims. Furthermore, structures typically found in information processing apparatuses / systems are not excluded. For example, the following variations 2 and onwards can be applied.

[0244] (Variation Example 2)

[0245] For example, in Figure 12 , Figure 19 The example shown illustrates the ability to set the scaling factor for both the X and Y directions, but it's also possible to set only one. Furthermore, it's also possible to set the scaling factor in the thickness direction. In this case, the X and Y directions are converted to the same scaling factor based on the thickness direction scaling factor setting input. For example, if the thickness direction scaling factor is 50%, the area becomes approximately 200%, and the X and Y scaling factors become 1 / 2 power of 144%.

[0246] (Variation Example 3)

[0247] Furthermore, when a batch is selected, if there are past results of products produced under the same production conditions, and a scaling factor set and determined by the user is recorded, the scaling factor applied in that past identical product can be read and displayed to the user. This allows the user to reduce the number of trials and errors, and thus easily set an appropriate scaling factor. Consequently, when using the same inspection device 90 between batches, if a deviation in coordinate setting occurs in the inspection device, it can be corrected.

[0248] (Variation Example 4)

[0249] In addition, Figure 19 The example shown illustrates the ability to divide the position along the length direction into multiple stages and set the stretching rate along the length direction for each division. However, it is also possible to set the stretching rate along the width direction for each division along the length direction. Similarly, when the position is also divided into multiple stages along the width direction, it is also possible to set the stretching rate along the length direction for each division along the width direction. In this way, for example, it is possible to correct the bowing of the membrane stretching, which is caused by the difference in the stretching rates along the length direction and / or the width direction at the position along the width direction.

[0250] (Variation Example 5)

[0251] Furthermore, the scaling factor can be inferred based on the degree of deformation of feature points and used as a recommended value to suggest to the user. For example, it can suggest... Figure 12 The value can be displayed in the middle or as the initial value for buttons b1 and b2.

[0252] (Variation Example 6)

[0253] Furthermore, for example, the information processing system 50 may also include an inspection device 90 configured for the first manufacturing process and / or the second manufacturing process. Additionally, the control unit 51 of the information processing system 50 may also perform the feature point generation function of the analysis unit 93 of the inspection device 90. In this case, image data obtained by photographing the film surface and its shooting conditions (transport speed, camera orientation, viewing angle, etc.) obtained from the inspection device 90 are sent to the information processing system 50, and feature point generation processing is performed on the control unit 51 side.

[0254] Furthermore, the units and methods for performing various processes in the information processing system 50 of the above embodiments can be implemented by either dedicated hardware circuitry or a programmed computer. The aforementioned programs can be provided, for example, via computer-readable recording media such as USB memory or DVD (Digital Multifunction Disc)-ROM, or online via a network such as the Internet. In this case, the programs recorded on the computer-readable recording media are typically transferred to a storage unit such as a hard disk and stored. Furthermore, the aforementioned programs can be provided as standalone application software, or they can be embedded in the software of the device as one of its functions.

[0255] Embodiments of the present invention have been described in detail and illustrated, but the disclosed embodiments are created for illustrative and exemplary purposes only and are not intended to be limiting. The scope of the invention should be interpreted by the wording of the appended claims.

[0256] This application is based on Japanese Patent Application No. 2024-056234, filed on March 29, 2024, the disclosure of which is incorporated herein by reference in its entirety.

[0257] Explanation of reference numerals in the attached figures

[0258] 50 Information Processing System

[0259] 51 Control Department

[0260] 511 Acquisition Department

[0261] 512 Acceptance Department

[0262] 513 Comparison Section

[0263] 514 Analysis Department

[0264] 515 Extraction Section

[0265] 516 Output Section

[0266] 52 Storage Unit

[0267] 90 Inspection device

[0268] 1000 film roller manufacturing equipment

[0269] 2000 Product Manufacturing Equipment

Claims

1. A method for extracting feature points of a roll material, comprising: Step (a) is to obtain first inspection data in the first manufacturing process of manufacturing roll material or performing post-processing on the manufactured roll material. Step (b) is to obtain second inspection data in a second manufacturing process that performs post-processing using the coil material after the first manufacturing process. Step (c) is to accept the input of the stretch rate of the roll produced from the first manufacturing process to the second manufacturing process. Step (d) involves comparing the first feature point information of the roll material in the first inspection data with the second feature point information of the roll material in the second inspection data using the accepted stretch rate; and Based on the comparison results of step (d), step (e) is to extract feature points that exist in both the first inspection data and the second inspection data, or feature points that exist in one of the inspection data and the second inspection data but not in the other inspection data.

2. The feature point extraction method according to claim 1, wherein, In step (e), feature points present in both the first inspection data and the second inspection data are extracted.

3. The feature point extraction method according to claim 1, wherein, The first inspection data and the second inspection data are inspection data obtained by processing the images or signals obtained from photographing the roll material to extract feature point information. The feature point information includes information about the location of the feature points on the roll material. Step (d) includes step (d1) of aligning the position of the roll material in the first inspection data and the second inspection data. In step (d1), a process is included that uses the scaling factor to move the position of a feature point of one of the first inspection data and the second inspection data.

4. The feature point extraction method according to claim 1, wherein, The elongation rate accepted in step (c) is the elongation rate in the width direction, the elongation rate in the length direction, and / or the elongation rate in the thickness direction of the roll material.

5. The feature point extraction method according to claim 3, wherein, The first inspection data and the second inspection data are inspection data obtained by processing the images or signals obtained from photographing the roll material to extract feature point information. The feature point information includes information about the size of the feature points and their position on the roll material. Step (d) includes step (d1) of aligning the position of the roll material in the first inspection data and the second inspection data. In step (d1), there is a process of moving the position of a feature point of one of the first inspection data and the second inspection data along the width direction and / or the length direction using the scaling factor in the width direction and / or the scaling factor in the length direction.

6. The feature point extraction method according to claim 2, wherein, Also includes: Step (f) calculates the deviation of the feature points extracted in step (e) from each other in the first and second inspection data; and Output the calculated deviation step (g).

7. The feature point extraction method according to claim 1, wherein, The stretch rate accepted in step (c) is the stretch rate of each section obtained by dividing the width direction of the roll into multiple stages.

8. A control program for causing a computer to execute the extraction method according to any one of claims 1 to 7.

9. An information processing system, comprising: The acquisition unit acquires first inspection data in a first manufacturing process of manufacturing a roll or performing post-processing on the manufactured roll, and second inspection data in a second manufacturing process of performing post-processing on the roll after the first manufacturing process. The receiving department receives inputs of the stretch rate of the roll material produced from the first manufacturing process to the second manufacturing process; The comparison unit uses the received stretch rate to compare the first feature point information of the roll material in the first inspection data with the second feature point information of the roll material in the second inspection data. as well as The extraction unit extracts feature points that exist in both the first inspection data and the second inspection data, or feature points that exist in one of the inspection data but not in the other, based on the comparison result of the comparison unit.

Images