Edge portion detection method for optical film

Abstract

In order to use an inspection method capable of accurately performing inspection of a fitting offset of a rectangular panel and an optical film, a detection method capable of reliably detecting an optical film edge of an optical film fitted to a rectangular panel is provided. The method includes: a conveying step of conveying a rectangular panel on which an optical film is layered; a photographing step of photographing an object region including an edge of the optical film on the rectangular panel a plurality of times using light from one or a plurality of light sources while changing a condition of the light irradiated to the rectangular panel; a false detection element exclusion step of excluding one or a plurality of elements that cause an element other than the edge to be erroneously detected as the edge in detecting the edge of the optical film from each of a plurality of images obtained by photographing the object region at a plurality of positions; a best image selection step of selecting a best image for detecting the edge from the plurality of images from which the one or a plurality of elements are excluded; and an edge detection step of detecting the edge using the best image.

Description

Technical Field

[0001] The present invention relates to a method for detecting the edge of an optical film used in the inspection of an optical display panel, and more specifically, to a method for detecting the edge of an optical film in which, when inspecting the adhesion offset between a rectangular panel and an optical film adhered thereto, the best image that can most reliably detect the edge of the optical film is selected from a plurality of captured images, and the edge of the optical film is detected using the best image. Background Technology

[0002] In optical display panels, various optical films with optical functions are laminated onto a rectangular panel as needed to achieve the display function. During the manufacturing process of the optical display panel, after the optical films are laminated onto the surface of the rectangular panel, an inspection of the lamination state between the rectangular panel and the optical films (so-called lamination offset inspection) is performed to confirm the lamination accuracy. Conventional lamination offset inspection methods include those described in Patent Documents 1 and 2.

[0003] Patent Document 1 discloses a method using a region camera and a linear camera configured to capture images of the corner of an optical display panel, to which a polarizer and a liquid crystal panel are bonded, from above a vertical distance. The image captured by the camera is used to calculate the bonding offset. Patent Document 2 discloses a method that, while transporting an optical display panel with an optical film bonded to an optical unit, captures images of the corner of the panel using a region sensor camera, thereby checking for bonding offset. In this method, the distance between the end of the optical unit and the end of the optical film is calculated based on the captured image, and the bonding offset is determined based on this distance.

[0004] Existing technical documents

[0005] Patent documents

[0006] Patent Document 1: Japanese Patent Application Publication No. 2011-197281

[0007] Patent Document 2: Japanese Patent Application Publication No. 2016-118580

[0008] Patent Document 3: Japanese Patent No. 4377964 Summary of the Invention

[0009] The problem that the invention will solve

[0010] In conventional bonding misalignment inspection methods, including Patent Documents 1 and 2, a specific area (hereinafter referred to as the target area) of the optical display panel, comprising the corners of both the rectangular panel and the optical film, enters the camera's shooting area, and an image is taken when it reaches the shooting point. The shooting point is typically the position where the edge of the optical film on the optical display panel (usually the foremost edge in the direction of movement of the optical film) reaches vertically below the camera. In the captured image, the edge of the optical film appears as a line (bright line) reflecting light from a light source at the edge of the optical film; by identifying this reflected light, the edge of the optical film within the image can be detected. However, depending on the state of the edge of the optical film bonded to the rectangular panel, the internal structure of the rectangular panel, and the state of reflection, it is sometimes difficult to detect the edge solely based on the image captured when the edge reaches the shooting point.

[0011] For example, if the edge of the optical film in the captured image is shaped to make it difficult to reflect light from the light source, or if the reflected light from the edge is difficult to reach the camera angle, it can be difficult to detect the edge of the optical film. Furthermore, in the captured image, if the bright line representing the edge of the optical film is difficult to distinguish from the bright line representing the edge of the rectangular panel (i.e., the outermost edge of the rectangular panel), the bright line representing the internal structure (e.g., the edge of the sealing material used to seal the liquid crystal in the liquid crystal panel, the wiring inside the rectangular panel, etc.), or if the bright line representing the adhesive stripes (hereinafter referred to as paste stripes) that sometimes appear when the rectangular panel and the optical film are first bonded together are brighter than the bright line representing the edge of the optical film, the actual position of the edge of the optical film may be misdetected, potentially resulting in a value different from the actual bonding offset.

[0012] The purpose of this invention is to provide a method for detecting the edge of an optical film, which can reliably detect the edge of an optical film attached to a rectangular panel, so as to provide an inspection method that can accurately perform the inspection of the attachment misalignment between the rectangular panel and the optical film.

[0013] Methods for solving problems

[0014] In this invention, for an optical display panel on which an optical film is stacked on a rectangular panel, multiple images of an object region containing the corners of the rectangular panel and the corners of the optical film are obtained using illumination light from a light source. Among these multiple images, the best image that can most reliably detect the edge of the optical film is selected, and the edge of the optical film is detected using the best image.

[0015] The invention is characterized in that, in each of a plurality of images captured, elements that would lead to false detection as the edge of the polarizing film are excluded from the image, and the plurality of images that have excluded them are compared, and the image that can more reliably detect the edge of the polarizing film is selected as the best image.

[0016] That is, the present invention provides an edge detection method for detecting the edge of an optical film stacked on a rectangular panel. The method includes: a conveying step of conveying the rectangular panel on which the optical film is stacked; an imaging step of taking multiple images of an object area containing the edge of the optical film on the rectangular panel by changing the conditions of the light illuminating the rectangular panel using light from one or more light sources; a false detection element exclusion step of excluding one or more elements from each of the multiple images obtained by taking multiple images of the object area, which would cause elements other than the edge to be falsely detected as the edge when detecting the edge of the optical film; an optimal image selection step of selecting the optimal image for edge detection from the multiple images from which one or more elements have been excluded; and an edge detection step of detecting the edge using the optimal image. Here, the conditions of the light illuminating the rectangular panel only need to be set such that the state of the edge of the optical film differs in each of the multiple images taken; for example, the angle and direction of the light illuminating the rectangular panel can be used. The method for changing the light conditions illuminating the rectangular panel is not limited. For example, it is possible to use light from multiple light sources configured at a certain height above the conveying surface of the rectangular panel, and to take pictures at multiple different positions during conveying using light from different light sources; or to use multiple light sources with different angles, directions, and amounts of light illuminating the rectangular panel, and to take multiple pictures at specific positions during conveying. Alternatively, it is also possible to change the light conditions illuminating the rectangular panel by using light from a single light source and taking pictures at multiple different positions during the conveying of the rectangular panel from the upstream side to the downstream side.

[0017] In one embodiment, the step of eliminating false detections may include: performing masking on one or more elements that generate false detections in each of a plurality of images, thereby eliminating one or more elements from the image. The one or more elements that generate false detections may be any one or a combination of the following: the edge of a rectangular panel, stripes generated in the adhesive bonding the rectangular panel to the optical film, the sealing material of the rectangular panel, and the internal wiring of the rectangular panel.

[0018] In one embodiment, the optimal image is preferably selected based on the brightness of the edge of the optical film after excluding one or more elements that would cause false detections from the image. More preferably, the optimal image is selected based on the brightness of multiple portions of the edge of the optical film after excluding one or more elements that would cause false detections from the image.

[0019] In one embodiment, the capture of multiple images is preferably performed while the rectangular panel is stopped at each capture. In another embodiment, the capture of multiple images is preferably performed using light from light sources disposed on the upstream and downstream sides of the transport direction, and more preferably using light from light sources disposed opposite each other in the width direction of the rectangular panel, in addition to these light sources.

[0020] Invention Effects

[0021] According to the present invention, since the best image that can reliably detect the bright line at the edge of the optical film is selected from a plurality of images that capture a specified area including the end of the optical film attached to the rectangular panel, the attachment misalignment of the optical film can be checked more easily and with high precision using this image. Attached Figure Description

[0022] Figure 1 This is a schematic diagram showing the configuration of an edge detection device used in LCD panel manufacturing to detect the edge of a polarizing film attached to a liquid crystal cell.

[0023] Figure 2 This is a flowchart illustrating the process of selecting the best image from multiple captured images in an optical edge detection method according to an embodiment of the present invention.

[0024] Figure 3 This is a flowchart illustrating a masking process for eliminating one or more elements that cause false detections when detecting the edge of a polarizing film, and a fractionation process for evaluating multiple images, in a method according to an embodiment of the present invention.

[0025] Figure 4 This image shows the masking process that removes bright lines from the edges of liquid crystal cells and paste stripes.

[0026] Figure 5 This is an image representing the masking process used to remove bright lines from the sealing material.

[0027] Figure 6 This is a flowchart illustrating the detailed process of fractionalization in a method according to an embodiment of the present invention.

[0028] Figure 7 It is an image representing the fractionalization process. Detailed Implementation

[0029] The following is a reference to the appendix. Figure 1 The embodiments of the present invention will be described in detail below.

[0030] In the following description, we will use the case of an optical display panel that uses a polarizing film as an optical film, uses a liquid crystal cell as a rectangular panel, and has an optical film bonded to the rectangular panel as a liquid crystal panel as an example. However, we are not limited to these cases. The present invention can also be used in the inspection of various optical display panels manufactured by bonding an optically functional film to a rectangular panel.

[0031] [Overview of LCD panel manufacturing apparatus and polarizing film edge inspection apparatus]

[0032] The edge detection method of the present invention can be used, for example, in an apparatus for manufacturing a liquid crystal panel by continuously bonding a polarizing film fed from a roller to a liquid crystal cell (an RTP-type apparatus). This method is used to detect the edge of the polarizing film in order to inspect the manufactured liquid crystal panel for instances where the polarizing film is bonded to the surface of the liquid crystal cell at a state deviating from a predetermined bonding position (hereinafter referred to as bonding offset). The RTP method involves continuously manufacturing a liquid crystal panel by sequentially peeling only the normal, defect-free sheet-shaped polarizing films, along with the adhesive layer, from a strip-shaped release film supported by multiple sheet-shaped polarizing films on a strip-shaped release film separated by an adhesive layer. These normal films are then bonded to the liquid crystal cell via the adhesive layer, thereby continuously manufacturing the liquid crystal panel. This continuous manufacturing system is distinguished from conventional individual bonding methods that bond pre-cut sheets of polarizing film to the liquid crystal cell and is called a "continuous bonding (RTP; roll-to-roll)" apparatus. For example, the apparatus described in Patent Document 3 can be used as an RTP-type apparatus.

[0033] Figure 1 This is a schematic diagram illustrating an example of the configuration of a polarizing film edge detection device for detecting the edge of a polarizing film bonded to a liquid crystal cell. This device can, for example, be assembled as part of an inspection process used in a liquid crystal cell manufactured using the RTP method described above, after the polarizing film has been bonded to it.

[0034] Figure 1The apparatus shown includes: a transport path 1 for transporting a liquid crystal panel P manufactured by bonding a polarizing film F to a liquid crystal cell C; an illumination 2 disposed above the transport path 1; and a camera 3 disposed above the illumination 2. When the target area A of the liquid crystal panel P transported by the transport path 1 is within the shooting range of the camera 3, the illumination 2 is illuminated multiple times from upstream to downstream of the transport direction D of the liquid crystal panel P, and multiple images Is of the target area A are captured by the camera 3. The target area A is the area used to check the bonding misalignment between the liquid crystal cell C and the polarizing film F. It is typically an area including the edge CE and corner CC of the liquid crystal cell C, and an area including the edge FE (which is the edge to be detected) and corner FC of the polarizing film F bonded to the liquid crystal cell C, but is not limited to this, as long as it includes at least the edge CE and FE. The multiple images Is of the target area A are sent to a general-purpose computer, where the computer selects the best image for detecting the edge FE of the polarizing film F, and the edge FE is detected using the selected best image.

[0035] As illumination, multiple illumination sources 2 configured to irradiate the object area A from different directions can be used. Preferably, the multiple illumination sources 2 are configured at least upstream and downstream of the transport direction D of the liquid crystal panel P. Figure 1 Illumination 21, 22), but in addition, it is preferred to also be arranged in the direction traversing the transport direction D (the width direction of the liquid crystal panel P) ( Figure 1 (Illumination 23, 24). By also arranging illumination 2 in the width direction of the liquid crystal panel P, the edge FS in the width direction of the polarizing film F can be detected more reliably, thus enabling higher precision bonding offset checks. Alternatively, illumination 2 can be, for example, a ring-shaped illumination that arranges multiple light sources in a circular pattern and allows each of these light sources to be illuminated individually. When using such a ring-shaped illumination, by individually illuminating multiple locations in the circumferential direction, light can be irradiated onto the target area A from multiple different directions. Alternatively, a single illumination source can be used instead of... Figure 1 The invention is implemented by using multiple lighting fixtures 21-24 as shown, which serve as lighting 2. The position of one lighting fixture 2 can be, for example, set as follows: Figure 1 The position of illumination 21 or illumination 22, etc. In this case, the transport of the liquid crystal panel P is stopped at multiple different positions during the transport of the liquid crystal panel P from the upstream side to the downstream side in the transport direction D, and illumination 2 is turned on when the transport is stopped to perform the shooting.

[0036] Depending on the purpose, the camera 3 used in this invention can be a linear camera, a zone camera, or the like. For example, when photographing a target area A while moving the liquid crystal panel P, a linear camera is preferred. When using a linear camera, it is also possible to photograph the target area A while moving the linear camera with the liquid crystal panel P stationary. As a method for photographing the target area A while moving the liquid crystal panel P, a zone camera can also be used, where the shutter speed is sufficiently faster than the moving speed of the liquid crystal panel P. A zone camera can also be used when the liquid crystal panel P is stationary while photographing the target area A.

[0037] In the following embodiment, a configuration is used where one illumination source 21 and 22 are respectively arranged upstream and downstream of the target area A in the transport direction D of the liquid crystal panel P, and two illumination sources 23 and 24 are arranged in the width direction of the liquid crystal panel P. Multiple images Is are captured, and the optimal image Ib for detecting the edge FE of the liquid crystal film F is selected from these images Is. Hereinafter, the present invention will be described as an embodiment using light from four illumination sources 21 to 24 arranged at a constant height from the transport surface of the liquid crystal panel P and with different illumination directions, and capturing images at multiple different positions during transport to change the conditions of light illuminating the liquid crystal panel P.

[0038] [Detection of the edge of the polarizing film]

[0039] (Transportation and photography of LCD panels)

[0040] exist Figure 1 In the illustrated apparatus, multiple images Is are captured on an object region A, including the edge CE and corner CC of the liquid crystal cell C and the edge FE and corner FC of the polarizing film F. A camera 3, configured to capture images of the object region A, is coordinated with the timing of the illumination of multiple illuminations 21 and 22 to acquire multiple images Is during the period from when the object region A enters the shooting range of the camera 3 until it leaves. Specifically, during the period from when the object region A enters the shooting range of the camera 3 until it leaves, the action of stopping the transport of the liquid crystal panel A and capturing images is repeated multiple times at multiple predetermined positions, thereby acquiring multiple images Is. These predetermined positions can be, for example, when the edge FE of the polarizing film F is upstream of the shooting point in the transport direction D, at the shooting point itself, or downstream of the shooting point in the transport direction D. The multiple illuminations 21 and 22 are preferably illuminated sequentially. The number of images Is acquired is not limited and can be determined by considering the required accuracy for detecting the edge FE of the polarizing film F and the processing speed for selecting the optimal image for detecting the edge FE.

[0041] The capture of multiple images Is is preferably performed by stopping the liquid crystal panel P, which moves from the upstream side to the downstream side, at each capture. However, this is not a limitation; the capture can also be performed sequentially while moving the liquid crystal panel P from the upstream to the downstream side of the capture range. When capturing images while moving the liquid crystal panel P, multiple images Is can be acquired by opening the shutter of the camera 3 at timed intervals, in conjunction with the illumination of multiple illuminations 21 and 22, until the object area A enters and leaves the field of view of the camera 3. In this case, the multiple illuminations 21 and 22 can be illuminated sequentially or simultaneously multiple times. Alternatively, multiple images Is can also be acquired by continuously capturing images by opening the shutter of the camera 3 while the object area A enters and leaves the field of view of the camera 3, during which time the multiple illuminations 21 and 22 are illuminated sequentially. The acquired multiple images Is are transmitted from the camera 3, for example, via wired or wireless means to a general-purpose computer (not shown), and stored in a storage unit such as a hard disk.

[0042] (Excluding false positives)

[0043] Multiple images Is are retrieved from the storage unit, and the best image Ib is selected from these images Is. The best image Ib is the image from the multiple images Is that can most reliably detect the edge FE of the polarizing film F attached to the liquid crystal cell C.

[0044] In this invention, the reason why bright lines in the captured image cannot be correctly detected as lines representing reflected light from the edge FE of the polarizing film F is based on the premise that bright lines representing elements other than the edge FE in the image are mistakenly detected as bright lines representing the edge FE. In this embodiment, as a reason why the edge FE cannot be detected, since the reflected light from the edge FE is relatively weak, in the image, examples include (1) a bright line representing the edge CE of the liquid crystal cell C existing on the outside of the polarizing film F is mistakenly detected as a bright line representing the edge FE, (2) a bright line representing the paste stripes existing on the inside of the polarizing film F is mistakenly detected as a bright line representing the edge FE, and (3) a bright line representing the edge of the sealing material existing on the inside of the polarizing film F is mistakenly detected as a line representing the edge FE.

[0045] Therefore, in this invention, other bright lines that might be misdetected as the edge FE during edge FE detection are excluded from each of the multiple captured images Is. The multiple images Is' from which these bright lines have been excluded are compared, and the image that can more reliably detect the edge FE is selected as the optimal image Ib. When one or more bright lines are excluded from image Is, the remaining bright lines in image Is' represent the edge FE of the polarizing film F. Therefore, by comparing the brightness of this bright line among the multiple images Is', the image with the higher brightness of the bright line can be selected as the optimal image Ib.

[0046] Figure 2 This is flowchart 200, which represents the overall process of selecting the best image Ib from multiple images Is. Additionally, Figure 3 This is a flowchart 300, which illustrates an example of masking processing to exclude one or more bright lines that cause false detections when detecting the edge EF, and fractional processing for evaluating multiple images Is. Figure 4 This image shows an example of a masking process that removes bright lines from the edge FE of the liquid crystal cell C and the paste stripes. Figure 5 This is an image illustrating an example of a masking process to remove bright lines of sealing material. Furthermore, Figure 6 Flowchart 400 is an example of a specific process in the fractionalization process. Figure 7 This is an image illustrating an example of a fractionalization process. See below for reference. Figures 2-7 This section specifically explains the method for selecting the optimal image Ib. Additionally, Figures 2-7 The optimal image selection process shown is just one example; other processes may be used depending on the type of optical display panel.

[0047] First, in S201 of flowchart 200, bright lines that were mistakenly detected as the edge FE of the polarizing film F are excluded from each of the multiple images Is retrieved from the storage unit. This yields multiple images Is' from which bright lines that caused false detection have been excluded. Next, in S202, for each of the multiple images Is', the brightness of the remaining bright lines after excluding the bright lines that caused false detection is measured, and each image Is' is evaluated based on its brightness. Next, in S203, the evaluation scores are compared among the multiple images Is', and the image with the highest evaluation score is selected as the optimal image Ib that can most reliably detect the edge FE.

[0048] Next, we will explain each process in flowchart 200 in detail.

[0049] In s201 of flowchart 200, in each of the plurality of images Is, bright lines that could cause false detections are excluded to obtain image Is'. In one embodiment, the exclusion of bright lines that could cause false detections can be achieved by applying a mask to the bright lines. For example, when the liquid crystal panel P is illuminated for inspection, in image Is, in addition to the bright lines representing the edge FE of the polarizing film F, there are also bright lines representing the edge CE of the liquid crystal cell C, bright lines representing the stripes (paste stripes) of the adhesive that sometimes forms when the liquid crystal cell C and the polarizing film F begin to adhere, and bright lines representing the edge of the sealing material used to seal the liquid crystal of the liquid crystal panel. Figure 3S301 to S307 illustrate the process used to obtain image Is' by performing masking processing on bright lines other than the bright lines representing the edge FE in each of the plurality of images Is. Furthermore, hereinafter, in image Is, the bright lines representing the edge CE of the liquid crystal cell C, the bright lines representing the paste stripes, and the bright lines representing the edge of the sealing material are sometimes simply referred to as the liquid crystal cell edge, the paste stripes, and the sealing material edge, respectively.

[0050] First, in s301, the edge of each liquid crystal cell is identified for each of the multiple images Is. The identification of the liquid crystal cell edge can be performed using pattern recognition technology. For example, since multiple wirings are implemented inside the edge PE of the liquid crystal panel P, the multiple wirings in the image Is are identified by pattern recognition, and the bright lines adjacent to the identified wirings can be regarded as the edge of the liquid crystal cell.

[0051] Pattern recognition can be performed using conventional methods. For example, an image of a small area containing the edges of liquid crystal cells and adjacent internal wiring of the liquid crystal panel P, which is the object of the alignment offset check, can be obtained in advance. Figure 4 Image Ir1 in (a) is registered, and in the captured image Is, regions with patterns consistent with the pre-registered image Ir1 are searched, and small regions are identified. Figure 4 (a) The area RR1 of the dashed line). Bright lines existing adjacent to the internal wiring in this small area are identified as bright lines representing the edge of the liquid crystal cell. In s302, a mask processing of a pre-specified size is performed on the bright lines representing the edge of the liquid crystal cell identified here. Figure 4 (b) The solid line area MK1) excludes the bright lines at the edge of the liquid crystal cell from the image.

[0052] Next, in s303, paste stripes are identified. The identification of paste stripes can be performed, for example, by the following method: First, multiple small regions containing the edges of the liquid crystal cells (which have already undergone masking) and the edges of the polarizing film and their surroundings are defined along the length direction of the liquid crystal cell edges. Figure 4 (c) shows a quadrilateral SR1), and the brightness within these small regions is measured and curved. The curve is as follows: Figure 4 As shown in (d), this can be represented as the relationship between the distance from the end of the small region toward the inside of the liquid crystal panel P and the brightness. The number and size of the small regions are not limited, as long as they are considered to be able to reliably identify the number and size of the bright lines of the paste stripes.

[0053] In each of the generated curves for multiple small regions, if there are two bright lines whose brightness exceeds a predetermined threshold (TH) for determining the presence or absence of the edge FE of the polarizing film F, the bright line (BL) closer to the inner side of the liquid crystal panel P is identified as a paste stripe. The brightness used as the threshold can be a preset value that allows reliable detection of the edge FE if the brightness is above this value. Furthermore, if there are three or more bright lines whose brightness exceeds the threshold, the second and subsequent bright lines from the outer side of the liquid crystal panel P towards the inner side are identified as paste stripes. In s304, a mask processing of a predetermined size is performed on the bright lines representing the identified paste stripes. Figure 4 The solid line region (MK2) of (e) is used to exclude the bright lines of the paste stripes from the image. Masking can be performed individually for each small region of bright lines, or a mask of the overall size can be implemented to exclude paste stripes from multiple small regions.

[0054] Furthermore, in s305, the sealing material edge is identified. The sealing material has a certain width in a direction parallel to the surface of the liquid crystal panel P. The edge identified here is the edge located in the inner direction of the liquid crystal panel P (otherwise, the edge located in the outer direction will not be identified as a bright line due to the contrast with the background, so no identification processing is required). The identification of the sealing material edge can also be performed using pattern recognition technology in the same way as the liquid crystal cell edge. For example, an image of a small area of ​​the liquid crystal panel P including the sealing material edge involved in the bonding offset check is obtained in advance ( Figure 5 Image Ir2 in (a) is registered, and a pattern matching the pre-registered image Ir2 is searched in the captured image Is to identify small regions ( Figure 5 (a) The area RR2 of the dashed line). The bright lines present in this small area are identified as the edge of the sealing material. In s306, a masking process of a pre-specified size is performed on the bright lines identified here as the edge of the sealing material. Figure 5 (b) The solid line area MK3) excludes the bright line of the sealing material edge from the image.

[0055] If bright lines representing the edges of liquid crystal cells, bright lines representing paste stripes, and bright lines representing the edges of sealing materials are excluded from the image, the remaining bright lines in the image are those representing the edges FE of the polarizing film F. This processing is performed on multiple captured images Is to obtain multiple images Is' (s307) that only show the bright lines representing the edges FE of the polarizing film F.

[0056] (Option selection)

[0057] Next, as shown in s202 of flowchart 200, for each of the acquired multiple images Is', an evaluation score is calculated to assess the likelihood of reliable detection of the edge FE of the polarizing film F. As shown in s203, the evaluation scores of the multiple images Is' are compared to select the optimal image Ib. Specifically, in s308 of flowchart 300, for each of the multiple images Is', the brightness is measured at multiple locations containing the edge FE. In s309, the brightness scores of the multiple locations are fractionated and summed, and the sum is used as the evaluation score. The image with the highest evaluation score among the multiple images Is' becomes the optimal image Ib capable of most reliably detecting the edge FE.

[0058] Figure 6 This is a flowchart 400 detailing an example of a process for calculating an evaluation score for each of multiple images Is', selecting the best image, and determining the position of the edge FE of the polarizing film F in each image Is'. First, in s401, multiple small regions including the edge of the polarizing film and its surroundings are defined along the length direction of the edge of the polarizing film. Figure 7 (a) shows a quadrilateral SR2). While the size of the small region is not limited, it is preferable to set the length of the small region in the direction traversing the edge of the polarizing film to include the edge of the liquid crystal cell after masking and the location of the paste stripes. Furthermore, it is preferable to appropriately set the length of the small region along the length direction of the polarizing film edge, taking into account the detection accuracy and processing speed of the edge. The number of small regions is not limited, but it is preferable to set it considering the detection accuracy and processing speed of the edge. Brightness is measured and curved in multiple small regions. The curve is shown below. Figure 7 As shown in (b), it can be expressed as the relationship between the distance from the end of the small region toward the inside of the liquid crystal panel P and the brightness.

[0059] Next, in s402, for each curve generated for multiple small regions, the intersection point of the line representing the curve and the predetermined threshold (TH) for determining the presence or absence of the edge FE of the polarizing film F is searched. The brightness used as the threshold is as described previously in the recognition of paste stripes, and can be assumed that if the brightness is above this value, the value of the edge FE can be reliably detected in the image. The bright line of the edge FE presented in the image has width, so there are usually two intersection points between the curve and the threshold. The position CP1 of the two intersection points, which corresponds to the intersection point in the inner direction of the liquid crystal panel P, is the position of the inner edge of the bright line representing the edge FE. If the positions of the inner edges of each small region thus identified are connected, the connecting line is the inner edge of the bright line representing the edge FE (s403), and this position becomes the position of the bright line representing the edge FE of the polarizing film F.

[0060] On the other hand, in s404, for each of the multiple small regions, the maximum luminance Bmax of the bright line representing the edge FE within that small region is fractionally divided. Fractionalization can be expressed, for example, as the relative luminance when the luminance corresponding to the maximum incident energy that the light-receiving element of camera 3 can accept is set to 100. Through luminance fractionalization, for example, a fraction for one small region is represented as 82, and fractions for another small region are represented as 85, 90, etc. Figure 7 (a)). After calculating the score for each of the multiple small regions, in s405, for example, the scores of all small regions are summed, and the sum is used as the evaluation score of the image. Furthermore, the evaluation score of the image is not limited to the sum of the scores of the small regions, as long as it can determine the detection reliability of the edges (FE) between images. For example, the evaluation score of the image can also be set as the average score of the scores of multiple small regions, the number of small regions with a score higher than a certain specific score, etc. The evaluation score is calculated in this way for each of the multiple images Is', and the image with the highest evaluation score is taken as the best image Ib.

[0061] To give a specific example, consider images A' and B' from two captured images, A and B, excluding bright lines other than the edge FE of the polarizing film F. Three small regions are defined along the edge FE, and the brightness of each small region is measured. If the scores of each small region in image A' are 82, 85, and 77, the total evaluation score for image A' is 244. Conversely, if the scores of each small region in image B' are 80, 95, and 50, the total evaluation score for image B' is 225. In this case, image A', with the higher evaluation score, is the optimal image Ib for more reliable detection of the polarizing film edge. Furthermore, the inner edge of the edge FE of the polarizing film F in image A' is a straight line connecting the inner edges of the three small regions.

[0062] The data generated in the processes of eliminating factors that cause false detections (processing flowchart 200, processing flowchart 3 ...

[0063] As described above, the optimal image Ib is selected from the multiple captured images Is to more accurately detect the edge FE of the polarizing film F attached to the liquid crystal cell C. The selected optimal image Ib can be used to detect the edge FE, for example, by determining the attachment offset of the polarizing film FE using methods known to those skilled in the art based on the relationship between the detected position of the edge FE and the position of the edge CE of the liquid crystal cell C.

[0064] The edge FE and its position in the selected optimal image Ib can be directly used from the image selected as the optimal image Ib, which were detected during the process of selecting the optimal image Ib. Alternatively, the selected optimal image Ib can be processed again with the same processing (masking) as that performed on each of the multiple images Is for selecting the optimal image Ib, and the finally detected edge FE and its position can be used as the edge and its position for determining the fitting offset.

Claims

1. A method for detecting the edge of an optical film, wherein the edge of an optical film laminated on a rectangular panel is detected, characterized in that, The optical film edge detection method includes: The conveying step involves conveying a rectangular panel with stacked optical films. The shooting process involves using light from one or more light sources to change the conditions of the light illuminating the rectangular panel and taking multiple shots of the object area containing the edge of the optical film on the rectangular panel. The false detection feature exclusion step excludes, from each of a plurality of images obtained by taking multiple images of the object region, one or more features that would cause the edge to be falsely detected as the edge when the edge is detected; The optimal image selection step involves selecting the optimal image for detecting the edge from among the plurality of images from which one or more features have been excluded; and The edge detection step uses the optimal image to detect the edge. The optimal image selection step includes: selecting the optimal image based on the brightness of the edge after excluding one or more elements from the image.

2. The optical film edge detection method according to claim 1, characterized in that, The false detection feature exclusion step includes: performing masking processing on one or more features in each of the plurality of images, thereby excluding one or more features from the image.

3. The optical film edge detection method according to claim 1 or 2, characterized in that, The one or more elements are any one of the following, or a combination thereof: the edge of the rectangular panel, the stripes generated in the adhesive that bonds the rectangular panel to the optical film, the sealing material of the rectangular panel, and the internal wiring of the rectangular panel.

4. The optical film edge detection method according to claim 1 or 2, characterized in that, The optimal image selection step includes selecting the optimal image based on the brightness of multiple locations defined along the edge after excluding one or more elements from the image.

5. The optical film edge detection method according to claim 1 or 2, characterized in that, The shooting steps include: stopping the rectangular panel during each shooting while capturing the multiple images.

6. The optical film edge detection method according to claim 1 or 2, characterized in that, The shooting step includes: capturing the plurality of images using light from light sources configured on the upstream and downstream sides of the conveying direction.

7. The optical film edge detection method according to claim 6, characterized in that, The shooting step includes: also capturing the plurality of images using light from light sources disposed opposite each other in the width direction of the rectangular panel.

Images