Image storage method and system

Abstract

An image storage method is provided. The method comprises the steps of: clustering images into clusters; extracting a common area between a plurality of images included in a first cluster included in the clusters; on the basis of the extracted common area, extracting a specialized area of each of the plurality of images included in the first cluster; and storing the common area and the specialized area of each of the plurality of images included in the first cluster.

Description

Image storage method and system

[0001] The present disclosure relates to an image storage method and system thereof. Specifically, the present disclosure relates to a method and system for storing images by separating common areas and specific areas of each image.

[0002] As technology advances, the manufacturing industry has also become capable of using computing systems to analyze images of products created during the manufacturing process to detect defects or manage manufacturing process designs, among other tasks.

[0003] For example, after a product is produced at a production base located overseas, an image of the produced product can be stored, and the image can be received and analyzed at a base located domestically to manage the manufacturing process performed overseas.

[0004] However, as the production volume of goods increases, the amount of images that need to be stored increases, which incurs a high cost for storing and managing a large number of images, and excessive network traffic may be generated in the process of transmitting a large number of images, which may result in reduced transmission efficiency.

[0005] Therefore, a new method is required to efficiently store and transmit large amounts of images.

[0006] The technical problem to be solved by the embodiments of the present disclosure is to provide a method for storing images and a computing system for performing the method. Specifically, the present disclosure provides a method for compressing and storing images and a computing system for performing the method.

[0007] The technical problem to be solved through other embodiments of the present disclosure is to provide a method for separating and extracting common areas of images and specialized areas of each image, and a computing system for performing the method.

[0008] A technical problem to be solved through other embodiments of the present disclosure is to provide a method for clustering images into clusters for efficient image storage and a computing system for performing the method.

[0009] Another technical problem to be solved by other embodiments of the present disclosure is to provide a method for extracting a common area where images overlap with a minimum of processing steps and a computing system for performing the method.

[0010] Another technical problem to be solved through embodiments of the present disclosure is to provide a method for transmitting an original image by separating an area of ​​the original image in order to reduce network delay / latency in the process of transmitting the image, and a computing system for performing the method.

[0011] Another technical problem to be solved by other embodiments of the present disclosure is to provide a method for restoring images stored in separated regions into original images and a computing system for performing the method.

[0012] The technical problems to be solved by the present disclosure are not limited to the technical problems mentioned above, and other technical problems not mentioned can be clearly understood by a person having ordinary skill in the art related to the present disclosure based on the detailed description below.

[0013] According to an embodiment of the present disclosure, a method performed by a computing system is provided. The method may include the steps of clustering images into clusters, extracting a common area between a plurality of images included in a first cluster included in the clusters, extracting a specialized area of ​​each of the plurality of images included in the first cluster based on the extracted common area, and storing the common area and the specialized area of ​​each of the plurality of images included in the first cluster.

[0014] In one embodiment of the present disclosure, the images are taken of each of a plurality of articles manufactured during a first time period, and the first time period includes a time point at which a design change related to the manufacture of the plurality of articles is reflected, and the step of clustering the images may include a step of dividing the first time period into a second time period and a third time period based on the time point at which the design change is reflected, and a step of clustering the images taken in the second time period into a second cluster, and a step of clustering the images taken in the third time period into a third cluster.

[0015] In one embodiment of the present disclosure, the step of clustering the images may include a step of calculating a similarity between the images using SSIM (Structed Similarity Image Matching), and a step of clustering the first image and the second image into one cluster when the similarity between the first image and the second image included in the images is greater than or equal to a preset threshold.

[0016] In one embodiment of the present disclosure, the step of extracting the common area between the plurality of images included in the first cluster may include the step of calculating the shortest path by applying a shortest path algorithm to the plurality of images based on the calculated similarity between the images, and the step of extracting the common area between images included in the calculated shortest path among the plurality of images.

[0017] In one embodiment of the present disclosure, the plurality of images included in the first cluster include a first image and a second image, and the step of extracting the common area between the plurality of images included in the first cluster may include the steps of extracting first feature points of the first image and second feature points of the second image, generating feature point pairs by matching each of the first feature points with each of the second feature points based on a similarity between each of the first feature points and each of the second feature points, and extracting a partial common area between the first image and the second image based on the feature point pairs.

[0018] In one embodiment of the present disclosure, the step of extracting the partial common area between the first image and the second image based on the feature point pairs may include the step of obtaining first pixel coordinates of the first image and second pixel coordinates of the second image, the step of transforming the second pixel coordinates of the second image so that the pixel coordinates of the feature point of the first image and the feature point of the second image included in each of the feature point pairs are identical, and the step of projecting the coordinate-transformed second image onto the first image to extract the partial common area.

[0019] In one embodiment of the present disclosure, the step of extracting the specialized region of each of the plurality of images included in the first cluster based on the extracted common region may include the steps of extracting first feature points of the extracted common region and second feature points of the first image included in the plurality of images, generating feature point pairs by matching each of the first feature points with each of the second feature points based on a similarity between each of the first feature points and each of the second feature points, obtaining first pixel coordinates of the extracted common region and second pixel coordinates of the first image, transforming the second pixel coordinates of the first image so that the pixel coordinates of the feature point of the extracted common region included in each of the feature point pairs and each of the feature points of the first image are identical, and projecting the coordinate-transformed first image onto the extracted common region to extract a first specialized region of the coordinate-transformed first image excluding the extracted common region.

[0020] In one embodiment of the present disclosure, the step of extracting the specialized region of each of the plurality of images included in the first cluster based on the extracted common region may further include the step of reconverting the transformed second pixel coordinates of the coordinate-transformed first image into the second pixel coordinates, and the step of obtaining third pixel coordinates of the first specialized region based on the second pixel coordinates of the first image.

[0021] In one embodiment of the present disclosure, for each of the clusters, a step of reducing the resolution of a plurality of images included in each cluster may be further included.

[0022] In one embodiment of the present disclosure, the first feature points and the second feature points can each be extracted based on Scale-Invariant-Feature Transform (SIFT).

[0023] According to another embodiment of the present disclosure, a system is provided. The system includes at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the at least one processor to perform operations, wherein the operations may include: clustering images into clusters; extracting a common area between a plurality of images included in a first cluster included in the clusters; extracting a specialized area of ​​each of the plurality of images included in the first cluster based on the extracted common area; and storing the common area and the specialized area of ​​each of the plurality of images included in the first cluster.

[0024] In one embodiment of the present disclosure, the images are taken of each of a plurality of articles manufactured during a first time period, and the first time period includes a time point at which a design change related to the manufacture of the plurality of articles is reflected, and the operation of clustering the images may include an operation of dividing the first time period into a second time period and a third time period based on the time point at which a design change is reflected, and an operation of clustering the images taken in the second time period into a second cluster and clustering the images taken in the third time period into a third cluster.

[0025] In one embodiment of the present disclosure, the operation of clustering the images may include an operation of calculating a similarity between the images using SSIM (Structed Similarity Image Matching), and an operation of clustering the first image and the second image into one cluster when the similarity between the first image and the second image included in the images is greater than or equal to a preset threshold.

[0026] In one embodiment of the present disclosure, the operation of extracting the common area between the plurality of images included in the first cluster may include an operation of calculating the shortest path by applying a shortest path algorithm to the plurality of images based on the calculated similarity between the images, and an operation of extracting the common area between images included in the calculated shortest path among the plurality of images.

[0027] In one embodiment of the present disclosure, the plurality of images included in the first cluster include a first image and a second image, and the operation of extracting the common area between the plurality of images included in the first cluster may include the operation of extracting first feature points of the first image and second feature points of the second image, the operation of matching each of the first feature points with each of the second feature points based on a similarity between each of the first feature points and each of the second feature points to generate feature point pairs, and the operation of extracting a partial common area between the first image and the second image based on the feature point pairs.

[0028] In one embodiment of the present disclosure, the operation of extracting the specialized region of each of the plurality of images included in the first cluster based on the extracted common region may include the operation of extracting first feature points of the extracted common region and second feature points of the first image included in the plurality of images, the operation of matching each of the first feature points with each of the second feature points based on a similarity between each of the first feature points and each of the second feature points to generate feature point pairs, the operation of obtaining first pixel coordinates of the extracted common region and second pixel coordinates of the first image, the operation of transforming the second pixel coordinates of the first image so that the pixel coordinates of the feature point of the extracted common region included in each of the feature point pairs and each of the feature points of the first image are identical, and the operation of projecting the coordinate-transformed first image onto the extracted common region to extract a first specialized region of the coordinate-transformed first image excluding the extracted common region.

[0029] According to another embodiment of the present disclosure, a computer program is provided. The computer program may be a program stored in a recording medium, coupled with a computing device, to execute the steps of clustering images into clusters, extracting a common area between a plurality of images included in a first cluster included in the clusters, extracting a specialized area of ​​each of the plurality of images included in the first cluster based on the extracted common area, and storing the common area and the specialized area of ​​each of the plurality of images included in the first cluster.

[0030] The above-described problem solving means are only some of the examples of the present disclosure, and various examples reflecting the technical features of the present disclosure can be clearly understood by a person having ordinary skill in the art related to the present disclosure based on the detailed description below.

[0031] Figure 1 is a configuration diagram of the entire system to which an image storage system according to one embodiment of the present disclosure is applied.

[0032] FIG. 2 is a flowchart illustrating an example of an image storage method according to one embodiment of the present disclosure.

[0033] FIGS. 3 and 4 are drawings illustrating specific examples of an image clustering process according to some embodiments of the present disclosure.

[0034] FIG. 5 is a flowchart illustrating a specific example of an image clustering process and a process of extracting common areas of images according to some embodiments of the present disclosure.

[0035] FIG. 6 is a diagram illustrating a process of extracting common areas of images by applying a shortest path algorithm according to some embodiments of the present disclosure.

[0036] FIG. 7 is a flowchart illustrating another specific example of a process for extracting common areas of images included in a cluster generated according to some embodiments of the present disclosure.

[0037] FIG. 8 is a diagram illustrating an example of images with feature points matched according to some embodiments of the present disclosure.

[0038] FIG. 9 is a flowchart illustrating a specific example of a process for extracting a specialized region of each image included in a cluster generated according to some embodiments of the present disclosure.

[0039] FIG. 10 is a diagram illustrating an example of a common area of ​​images included in a cluster generated according to some embodiments of the present disclosure and a specialized area of ​​each image.

[0040] FIG. 11 is a flowchart illustrating an example of the overall operation process of an image storage system according to some embodiments of the present disclosure.

[0041] FIG. 12 is a block diagram illustrating an example of a computing device for performing some embodiments of the present disclosure.

[0042] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the attached drawings. The advantages and features of the present disclosure, and methods for achieving them, will become clearer with reference to the embodiments described in detail below together with the attached drawings. However, the present disclosure is not limited to the embodiments described below and may be implemented in various different forms. These embodiments are provided only to ensure that the present disclosure is complete and to fully inform those skilled in the art of the present disclosure of the scope of the invention, and the embodiments of the present disclosure are defined only by the scope of the claims.

[0043] To avoid ambiguity in the concepts of this disclosure, the known components may be omitted or illustrated in block diagram form focusing on the core functions of each component. Throughout this disclosure, identical components are described using the same reference numerals, even if they appear on different drawings.

[0044] Unless otherwise defined, all terms (including technical and scientific terms) used herein may be used in the same sense as commonly understood by those of ordinary skill in the art to which this disclosure pertains. Furthermore, terms defined in commonly used dictionaries are not to be interpreted ideally or excessively unless explicitly and specifically defined otherwise. The terminology used herein is for the purpose of describing embodiments and is not intended to limit the present invention. In this disclosure, singular forms also include plural forms, unless specifically stated otherwise.

[0045] In addition, the terminology used in this disclosure is only used to describe specific embodiments, and is not intended to limit the features, components, order, etc. described in the specification. The terms "comprises" and / or "comprising" used in this disclosure express the presence of features, components, steps, operations, and / or combinations thereof described in the specification, but do not exclude the presence or addition of one or more other features, components, steps, operations, and / or combinations thereof.

[0046] Additionally, terms such as first, second, A, B, (a), (b), etc. used in the following embodiments are only used to distinguish certain components from other components, and the nature, order, or sequence of the components are not limited by the terms. For example, without departing from the scope of the present disclosure, the first component may be referred to as the second component, and similarly, the second component may be referred to as the first component.

[0047] Additionally, in this disclosure, " / " and "," should be interpreted as indicating "and / or". For example, "A / B" and "A, B" may mean "A and / or B".

[0048] The present disclosure proposes a method and system for storing images. Specifically, the present disclosure proposes a method and system for compressing and storing a plurality of images.

[0049] For example, according to some embodiments of the present disclosure, common areas of images and specialized areas of each image can be extracted, and the common areas and specialized areas of each image can be stored separately. For another example, according to some embodiments of the present disclosure, images can be clustered into clusters based on a predetermined criterion.

[0050] Hereinafter, specific embodiments will be described with reference to the drawings.

[0051] FIG. 1 is a configuration diagram of the entire system to which an image storage system (300) according to one embodiment of the present disclosure is applied.

[0052] Referring to FIG. 1, the overall system may include a first user device (100), a second user device (200), and / or an image storage system (300).

[0053] The overall system of FIG. 1 may provide a framework for compressing and storing images, or restoring compressed images for use in image analysis, according to some embodiments of the present disclosure. For example, the overall system may provide a framework for integrated management of a manufacturing process, such as photographing products produced during the manufacturing process and storing / transmitting / receiving / analyzing the acquired images.

[0054] The first user device (100) and / or the second user device (200) may include various devices that a user uses to communicate and transmit and receive various data and / or information with other devices. The first user device (100) and / or the second user device (200) may include, but is not limited to, a smartphone, a tablet PC, a laptop, etc. For example, the first user device (100) and / or the second user device (200) may include various computing devices equipped with a wireless communication means and / or a calculation means. The first user device (100) and / or the second user device (200) may be referred to as a user terminal, a wireless device, a mobile terminal, a portable device, etc.

[0055] In the present disclosure, a user may refer to a person who stores / analyzes and manages compressed images according to some embodiments of the present disclosure. For example, a user may refer to a person who analyzes and / or manages images compressed and stored by an image storage system (300) using a first user device (100) and / or a second user device (200).

[0056] The first user device (100) and / or the second user device (200) may communicate and transmit data and / or information acquired / generated according to some embodiments of the present disclosure to and from other devices. For example, the first user device (100) may compress product images acquired during a manufacturing process through an image storage system (300) and transmit the compressed images to the second user device (200). For example, the second user device (200) may correspond to a device used by a user to restore compressed images to original images according to some embodiments of the present disclosure.

[0057] For reference, the images in this disclosure are described by way of example, as images taken of items produced during a manufacturing process. Here, the items may include components (e.g., circuit boards, etc.) that may be components of other items / machines / devices. However, this is merely an example for explaining embodiments of the present disclosure and is not limiting. Except for embodiments that are applicable only to images taken of items produced during a manufacturing process, the embodiments in this disclosure are applicable to all forms of images that include one or more objects, regardless of type, such as a person, body, animal, or plant.

[0058] For example, the first user device (100) and / or the second user device (200) may correspond to various devices used by users to manage the manufacturing process based on images taken of products produced during the manufacturing process.

[0059] Additionally, the first user device (100) and / or the second user device (200) may display a user interface in which the functions of the image storage system (300) are implemented according to some embodiments of the present disclosure.

[0060] The image storage system (300) is a computing system for performing methods and / or operations according to some embodiments of the present disclosure.

[0061] The image storage system (300) can compress images and store the compressed images according to some embodiments of the present disclosure. For example, the image storage system can cluster a large number of original images into clusters based on a predetermined criterion, compress the original images contained in each cluster, and store them for each cluster.

[0062] In the present disclosure, compressing an image may mean extracting a portion of the entire area of ​​an image. Accordingly, a compressed image in the present disclosure may mean a portion of the entire area of ​​an original image extracted from the original image. For example, in order to store a large number of images, the image storage system (300) may extract a common area between original images and extract a specific area of ​​each original image, and as compressed images, store or transmit only the extracted common area and the specific area of ​​each original image to another device.

[0063] The common area of ​​the images refers to the overlapping area between the images, and the specialized area of ​​each image may refer to the area of ​​the entire area of ​​each image excluding the common area. Accordingly, the specialized area of ​​the image may include a first specialized area and a second specialized area corresponding to different pixel coordinates.

[0064] Additionally, the image storage system (300) may restore a compressed image to an original image according to some embodiments of the present disclosure.

[0065] Although not illustrated in FIG. 1, the image storage system (300) may include a database for storing data acquired / generated according to some embodiments of the present disclosure. For example, the image storage system (300) may include a database for storing common areas and / or specialized areas extracted from an original image.

[0066] The image storage system (300) may be implemented on at least one computing device. For example, all functions of the image storage system (300) may be implemented on a single computing device. In another example, some functions of the image storage system (300) may be implemented on a first computing device, and the remaining functions may be implemented on a second computing device. Furthermore, specific functions of the image storage system (300) may be implemented on one or more computing devices.

[0067] Additionally, although not illustrated in FIG. 1, the overall system of FIG. 1 may further include a server for analyzing and / or restoring images compressed by the image storage system (300). For example, the image storage system (300) may transmit compressed images to this server according to some embodiments of the present disclosure. That is, the server included in the overall system may receive common areas between images and specialized areas of each image from the image storage system (300).

[0068] For example, a server that receives common areas between images and specialized areas of each image can restore original images based on the common areas and specialized areas of each image according to some embodiments of the present disclosure.

[0069] Additionally, the server can analyze compressed images or reconstruct the original images and analyze the reconstructed images. For example, the server can analyze compressed images and / or reconstructed original images for anomaly detection. The server may also utilize one or more AI models for image analysis.

[0070] The components included in the overall system of FIG. 1 can communicate via various types of wired / wireless networks. The device and / or system according to the present disclosure can be applied to a local area network (LAN), a wide area network (WAN), a mobile radio communication network, Wibro (Wireless Broadband Internet), etc., and is not limited thereto, and can also be applied to any other communication system.

[0071] Hereinafter, with reference to FIGS. 2 to 11, embodiments in which a computing system performs operations for image compression and / or storage according to embodiments of the present disclosure will be described in detail. For reference, FIGS. 2, 3, 5, 7, 9, and 11 illustrate steps / operations performed in the image storage system (300) of FIG. 1. Therefore, in the following description, if the subject of a specific step / operation is omitted, it can be understood that the step / operation is performed in the image storage system (300) of FIG. 1. Hereinafter, embodiments of the present disclosure will be described with reference to FIG. 1 together with FIGS. 2 to 11.

[0072] First, FIG. 2 is a flowchart illustrating an example of an image storage method according to one embodiment of the present disclosure.

[0073] Referring to FIG. 2, images can be clustered into clusters (S100). For example, images can be clustered based on image creation date / time, structural characteristics between images, brightness, contrast, etc. However, this is not limited thereto. For example, if the images are photographs of products produced during a manufacturing process, the images can be clustered into clusters based on the product's production line, product type, manufacturing company, anomaly / defect location, product code, and inspection device corresponding to each product. Embodiments related to criteria for image clustering are described in detail below with reference to FIGS. 3 and 4.

[0074] As the amount of images increases, the amount of resources required for storage / transmission / recovery may increase, and the load on the system may increase. However, according to some embodiments of the present disclosure, the time resources required for storing / transmitting / recovering images may be reduced by clustering the images into clusters and storing / transmitting / recovering the images for each cluster. In addition, since a multi-threaded configuration is possible during the storage / recovery process for a large number of images, the consumption of space resources and system resources may be reduced, and the load on the system may be reduced.

[0075] Furthermore, by clustering images into clusters in S100, common areas and specific areas of each image can be extracted for each cluster. This can increase image compression and efficiently utilize finite space resources.

[0076] Although not shown in FIG. 1, the resolution of the plurality of images included in each of the clusters generated in S100 may be reduced. In other words, the image storage system (300) may compress the images by reducing the resolution of the images for each cluster. For example, the image format of the plurality of images included in each of the clusters generated in S100 may be converted to AVIF (AV1 Image File Format) based on AV1 (Alliance for Open Media Video 1), thereby reducing the size of the image files for each cluster.

[0077] By reducing the resolution of images for each cluster, common areas between multiple images within the same cluster can be precisely and quickly extracted. In this disclosure, the process of reducing the resolution of images can be referred to as the image flattening process.

[0078] Below, the process of compressing images by cluster is described in detail. For convenience of explanation, below, the process of compressing images included in the first cluster included in the clusters in S100 is described, and it should be noted that steps S200 to S400 can be applied to each cluster included in the clusters in S100 or to images included in each cluster.

[0079] First, a common area among multiple images included in a first cluster can be extracted (S200), and based on the common area extracted in S200, a specialized area of ​​each of the multiple images included in the first cluster can be extracted (S300). In other words, the multiple images included in the first cluster can be compressed into a common area and a specialized area of ​​each image.

[0080] In S200, a common area between multiple images means an overlapping image area between the multiple images, and even if there is a same image area between the multiple images, if the area is located at different locations on different images, it may not be included in the common area. For example, even if a first cluster includes a first image and a second image, and a pixel value corresponding to a first area of ​​the first image and a pixel value corresponding to a second area of ​​the second image are the same, if the pixel coordinates corresponding to the first area and the second area are different, the areas corresponding to the first area and the second area may not be included in the common area.

[0081] In S300, the specialized region of each of the plurality of images may refer to an image region excluding a common region among the entire regions of each of the plurality of images. Accordingly, the specialized region of a first image included in a first cluster and the specialized region of a second image included in the first cluster and different from the first image may be different or identical.

[0082] The common region extracted from S300 and the specialized region of each of the multiple images included in the first cluster can be stored (S400). In other words, the multiple images included in the first cluster can be stored separately into the common region and the specialized region of each image.

[0083] According to embodiments of the present disclosure, a plurality of images included in a single cluster can be stored separately as one image, which is a unified image area as a common area among the plurality of images, and as a specialized area of ​​each of the plurality of images, so that the common area and the specialized areas of each of the plurality of images can be managed independently.

[0084] By storing the common area of ​​multiple images and the specialized area of ​​each of the multiple images separately, the efficiency of storage space use can be increased compared to storing original images with relatively large image data capacity, including the common area.

[0085] In addition, according to some embodiments of the present disclosure, in the process of transmitting and receiving a plurality of images, a common area of ​​the plurality of images and a specialized area of ​​each of the plurality of images can be transmitted and received separately, so that network traffic can be reduced compared to when original images are transmitted and received.

[0086] Next, the image clustering process is specifically described with reference to FIGS. 3 and 4.

[0087] Specifically, FIG. 3 is a flowchart illustrating a specific example of a process of clustering images into clusters in S100 of FIG. 2. S100 of FIG. 3 may correspond to S100 of FIG. 2.

[0088] Referring to Figure 3, images can be clustered into clusters based on the time point at which design changes are reflected.

[0089] For example, the images may be taken of each of a plurality of items manufactured during a first time interval, and the first time interval may include a time point at which a design change related to the manufacture of the plurality of items is reflected. In this case, the first time interval may be divided into a second time interval and a third time interval based on the time point at which the design change is reflected (S110), and the images taken during the second time interval may be clustered into a second cluster, and the images taken during the third time interval may be clustered into a third cluster (S120).

[0090] In other words, images can be clustered into different clusters based on the point in time when design changes are reflected.

[0091] Referring to Fig. 4, an embodiment of clustering images based on the time point at which design changes are reflected is specifically described.

[0092] Figure 4 is a diagram for explaining clustering images into clusters based on the time / time interval at which design changes are reflected.

[0093] Referring to 4A of FIG. 4, images whose creation time / period is before the time of reflection of the design change and images whose creation time / period is after the time of reflection of the design change can be clustered into different clusters.

[0094] Throughout the manufacturing process, design changes may be continuously made to improve product performance, reduce costs, and detect / manage defects. These design changes may introduce new defects / defects into the product, unlike those before the design change.

[0095] For example, as illustrated in 4A, if images are generated during a first time interval, and images generated before the design change reflection time are clustered into cluster 1, and images generated after the design change reflection time are clustered into cluster 2, the ratio of common areas among all images generated during the first time interval may be lower than the ratio of common areas among images included in cluster 1 and / or cluster 2.

[0096] Accordingly, according to some embodiments of the present disclosure, images of items created based on the time point of reflecting a design change are clustered, and the images are compressed by cluster, thereby increasing the image compression ratio.

[0097] Referring to Figure 4B, multiple design changes may be reflected during the manufacturing process, and the time intervals over which each design change is reflected may be the same, different, or overlap only for a portion of the time interval. In this case, images may be clustered into clusters based on the start and / or end points of the design change reflection time interval.

[0098] For example, as illustrated in 4B, if images are generated during a first time interval, and design changes 1, 2, and 3 are reflected during the first time interval, and the time intervals in which design changes 1 to 3 are reflected partially overlap each other, the images generated during the time interval from the start time of reflection of design change 1 to the start time of reflection of design change 2 may be clustered into cluster 1, the images generated during the time interval from the start time of reflection of design change 2 to the end time of reflection of design change 1 may be clustered into cluster 2, the images generated during the time interval in which only design change 2 is reflected may be clustered into cluster 3, the images generated during the time interval from the start time of reflection of design change 3 to the end time of reflection of design change 2 may be clustered into cluster 4, and the images generated during the time interval from the end time of reflection of design change 2 to the end time of reflection of design change 3 may be clustered into cluster 4.

[0099] The embodiments described with reference to FIGS. 3 and 4 are examples of clustering images into clusters for efficient storage / transmission / recovery of images, but are not limited thereto. For example, images of items produced during the manufacturing process can be clustered based on the product's production line, product type, manufacturing company, anomaly / defect location, product code, and the inspection device corresponding to each item.

[0100] During the manufacturing process, products may contain defects / flaws (hereinafter, "defects"). These defects may be symmetrical or diagonal. Defects may include genuine defects, which refer to defects in the product caused by issues during the actual manufacturing process, and potential defects, which refer to defects in the product that prevent it from functioning properly or cause errors during operation, even if the product appears to be normal on the outside.

[0101] For example, an image can be virtually divided into four square regions based on the center, and images with defects detected in the region located in the lower right can be clustered into the first cluster, images with defects detected in the region located in the lower left can be clustered into the second cluster, images with defects detected in the region located in the upper left can be clustered into the third cluster, and images with defects detected in the region located in the upper right can be clustered into the fourth cluster.

[0102] When clustering images into clusters based on the defective pattern / location of the product, the defective pattern / location detected in each image within the same cluster is more likely to be identical. Therefore, in this case, the proportion of common areas between images within the same cluster may be higher than when clustering is not based on the defective pattern / location.

[0103] Additionally, items produced during the manufacturing process may be components (e.g., circuit boards) that can become components of other items / machines / devices, and may be assigned an item code. Here, the item code may be a unique code for the finished product containing the item, and may be identical to or different from the item codes of other items. In other words, the item codes for the items corresponding to each image may be identical or different.

[0104] For example, an item code may be a purchase order number, which may indicate that items and / or finished goods with the same purchase order number were created at a predetermined time with the same specifications and quality characteristics.

[0105] Therefore, when clustering images into clusters based on product codes, the locations of defects and other features detected within each image within the same cluster are likely to be consistent. In this case, the proportion of common areas among images within the same cluster may be higher than when images are not clustered based on product codes.

[0106] Therefore, according to some embodiments of the present disclosure, the image compression ratio can be increased by clustering images into clusters based on defective patterns / locations, product codes, etc., and compressing the images by cluster.

[0107] Next, with reference to FIG. 5, a process of extracting common areas of images included in clusters generated according to some embodiments of the present disclosure will be described in detail.

[0108] Specifically, FIG. 5 is a flowchart illustrating a specific example of processes S100 to S200 of FIG. 2. S100 and S200 of FIG. 5 may correspond to S100 and S200 of FIG. 2, respectively.

[0109] Referring to FIG. 5, in S100, similarity between images can be calculated using SSIM (Structed Similarity Image Matching) (S130), and if the similarity between the first image and the second image included in the images is greater than a preset threshold value, the first image and the second image can be clustered into one cluster (S140).

[0110] SSIM is one of the methods for calculating the similarity between two images by considering three elements of the image structure, brightness, and contrast. According to some embodiments of the present disclosure, rather than simply clustering images based on differences in pixel values, the image compression ratio can be increased by grouping images with high similarity into one cluster by comparing the structural features of the images.

[0111] In addition, in S200, a shortest path algorithm can be applied to a plurality of images included in the first cluster based on the similarity between the images produced in S130 to produce a shortest path (S210), and a common area between images included in the shortest path produced among the plurality of images included in the first cluster can be extracted (S220).

[0112] In other words, the common area between the plurality of images included in the first cluster produced in S200 is extracted by comparing two images included in the plurality of images according to the shortest path produced in S220, and the common area between the two images is extracted, and the common area in S200 can be extracted by repeatedly performing this for the plurality of images included in the first cluster according to the shortest path. For example, in S130, the similarity between the plurality of images included in the first cluster is produced, and in S210, the shortest path algorithm can be applied using the similarity between the plurality of images included in the first cluster as a weight. A specific embodiment thereof will be described in detail below together with the description of FIG. 6.

[0113] The shortest path algorithm can be either Dijkstra's algorithm or Bellman-Ford algorithm.

[0114] According to some embodiments of the present disclosure, in order to store a large number of images, common areas between images and specialized areas of each image can be extracted, and the common areas and specialized areas of each image can be stored separately. The common areas between images can be extracted by comparing each image with each other. In this case, when only images included in a shortest path calculated based on similarity between images are compared and the common areas are extracted, the common areas can be extracted quickly with a smaller number of comparisons, rather than when all images within a cluster from which the common areas are to be extracted are compared with each other and the common areas are extracted.

[0115] The embodiments of the present disclosure may be performed individually or in combination. For example, images may be automatically classified based on the image generation date / time, the product's production line, the product type, the production company, the location of anomalies / defects, the product code, the inspection device corresponding to each product, etc., and the classified images may be clustered into clusters based on SSIM image similarity.

[0116] Therefore, according to some embodiments of the present disclosure, common areas between images can be extracted using a shortest path algorithm, thereby reducing system resource consumption and increasing computing processing speed for common area extraction.

[0117] Next, referring to FIG. 6, a process of calculating the shortest path by applying a shortest path algorithm to multiple images included in the first cluster based on the similarity between the images produced in S130 is described in detail.

[0118] Specifically, FIG. 6 is a diagram for explaining the process of extracting common areas of images by applying the Dijkstra algorithm.

[0119] For example, if a plurality of images included in the first cluster include a total of 7 images, images 0 to 6, a graph as illustrated in FIG. 6 can be constructed by making images 0 to 6 each a node and using the similarity values ​​of images 0 to 6 each as the weights of the edges. In other words, nodes 0 to 6 in the graph illustrated in FIG. 6 represent images included in the same cluster, and the weights of the edges connecting the two nodes can represent the similarity values ​​of the two nodes included in each of the two nodes. For example, in the example illustrated in FIG. 6, the similarity between images 0 and 4 is 3, and the similarity between images 0 and 5 is 10.

[0120] In the example shown in Fig. 6, if Dijkstra's algorithm is applied to the graph shown in Fig. 6 to derive a path that minimizes the sum of edge values ​​from node 0 to node 6, as shown in Fig. 6,

[0121] When Dijkstra's algorithm is applied to the graph illustrated in Fig. 6, a path can be derived that minimizes the sum of edges from node 0 to node 6. Referring to Fig. 6, in this example, a shortest path can be derived in which edges are indicated by arrows as illustrated in Fig. 6.

[0122] For example, in the example of FIG. 6, there is a method of extracting the common area between image 0 and image 1 by comparing image 0 and image 1, sequentially extracting the first common area between image 0 and image 4 according to the shortest path, and extracting the second common area between image 4 and image 1 to extract the final common area included in both the first common area and the second common area.

[0123] The edge / similarity value between node 0 and node 1 is 7, and the sum of the edge / similarity values ​​of node 0, node 4, and node 1 is 5. When image 0 and image 4 are sequentially compared according to the shortest path, and image 4 and image 1 are compared to extract the common area between image 0 and image 1, the common area between image 0 and image 1 may be a narrower range of image areas than the common area between image 0 and image 1 extracted by comparing image 0 and image 1 all at once without following the shortest path. In addition, when extracting the common area between image 0 and image 1 according to the shortest path, the number of comparisons may increase compared to the case where it is not.

[0124] However, the probability that the common region between multiple images (images 0 to 6) included in the same cluster includes the common region between image 0 and image 1 extracted along the shortest path is higher than the probability that the common region between image 0 and image 1 extracted without following the shortest path is included. Therefore, comparing image 0 and image 1 is an unnecessary comparison operation both for extracting the common region between image 0 and image 1 and for extracting the common region between image 0 to image 6.

[0125] Additionally, from the perspective of extracting common areas of multiple images (image 0 to image 6) included in the same cluster, the number of comparisons between the total images can be reduced compared to comparing all images individually without following the shortest path.

[0126] That is, according to some embodiments of the present disclosure, a shortest path algorithm is applied to a plurality of images included in a cluster to derive a shortest path, and a common area between the images is extracted according to the shortest path, so that the final common area between the plurality of images included in the cluster can be efficiently extracted with a minimum number of comparisons.

[0127] Next, another specific example of a process for extracting common areas of images included in a cluster generated according to some embodiments of the present disclosure is described with reference to FIG. 7.

[0128] Specifically, FIG. 7 is a flowchart illustrating an example of a process for extracting a common area between a first image and a second image included in a plurality of images included in a first cluster in order to extract a common area in S200 of FIG. 2. S200 of FIG. 7 may correspond to S200 of FIG. 2.

[0129] For reference, in the present disclosure, a common area between two images included in the extracted multiple images may be called a partial common area in order to extract a common area between multiple images.

[0130] Referring to FIG. 7, in S200, first feature points of a first image and second feature points of a second image of a second image are extracted (S201), and based on the similarity between each of the first feature points and each of the second feature points, each of the first feature points and each of the second feature points can be matched to generate feature point pairs (S202).

[0131] In the present disclosure, feature points of an image may be named feature descriptors.

[0132] In S201, the first and second feature points can be extracted based on Scale-Invariant-Feature Transform (SIFT), respectively. By extracting feature points from images based on SIFT, consistent feature points can be extracted for each of the multiple images included in the same cluster, even if each of the images has different angles or sizes.

[0133] Based on the feature point pairs generated in S202, a partial common area between the first image and the second image can be extracted. Here, in order to extract the partial common area between the first image and the second image, first pixel coordinates of pixels corresponding to the first image and second pixel coordinates of pixels corresponding to the second image are acquired (S203), and the second pixel coordinates of the second image can be converted so that the pixel coordinates of each of the feature points of the first image and the feature points of the second image included in each of the feature point pairs are identical (S204).

[0134] Additionally, the second image whose coordinates have been transformed in S204 can be projected onto the first image so that a partial common area between the first image and the second image can be extracted.

[0135] The implementations of S201 to S205 may be repeatedly performed on two images of a plurality of images included in a first cluster, thereby extracting a common region between the final plurality of images. For example, the implementations of S201 to S205 may be performed on two images included in the plurality of images according to the shortest path derived by applying a shortest path algorithm to the plurality of images included in the first cluster, thereby extracting a partial common region between the two images.

[0136] For example, the image storage system (200) can match feature points of a first image with feature points of second images, and calculate a homography matrix value using the coordinates of the feature points of each of the first image and / or the second image to extract a partial common area between the first image and the second image. That is, the common area can be extracted using the homography matrix. A specific description thereof will be described below with reference to FIG. 8.

[0137] FIG. 8 is a diagram illustrating an example of images with feature points matched according to some embodiments of the present disclosure.

[0138] Specifically, the process of homography-converting an image (90) into a reference image (80) is described with reference to FIG. 8.

[0139] Referring to FIG. 8A, according to some embodiments of the present disclosure, a reference image (80) and an image (90) may be included in the same cluster, and feature points (81) of the reference image (80) may be extracted, and feature points (91) of the image (90) may be extracted. As illustrated in FIG. 8, feature point pairs (81, 91) may be generated based on the similarity between the feature points of each of the two images.

[0140] Additionally, according to some embodiments of the present disclosure, an image (90) may be homographically transformed with respect to the reference image (80) to extract a partial common area between the reference image (80) and the image (90), as illustrated in 8B of FIG. 8. In this case, the image (90) may be projected onto the reference image (80) to extract a common area between the two images.

[0141] Next, a process of extracting a specialized region of each image included in a cluster generated according to some embodiments of the present disclosure will be described in detail with reference to FIG. 9.

[0142] Specifically, FIG. 9 is a flowchart illustrating an example of a process for extracting a specialized region of each of a plurality of images included in a first cluster in S300 of FIG. 2. S300 of FIG. 9 may correspond to S300 of FIG. 2.

[0143] Referring to FIG. 9, in S300, first feature points of the extracted common area and second feature points of the first image included in the plurality of images included in the first cluster are extracted (S301), and feature point pairs can be generated by matching each of the first feature points with each of the second feature points based on the similarity between each of the first feature points and each of the second feature points (S302).

[0144] In addition, the first pixel coordinates of the extracted common area and the second pixel coordinates of the first image are obtained (S303), and the second pixel coordinates of the first image can be transformed so that the pixel coordinates of the feature points of the common area included in each of the feature point pairs generated in S302 and each of the feature points of the first image are identical (S304), and the coordinate-transformed first image can be projected onto the extracted common area so that a first specialized area, which is an area excluding the extracted common area on the coordinate-transformed first image, can be extracted (S305).

[0145] In S301, the first feature points and the second feature points can each be extracted based on SIFT (Scale-Invariant-Feature Transform).

[0146] For example, in S301 to S305, the common area and the feature points of each of the first images are matched to project the first image onto the common area to extract the specialized area of ​​each of the plurality of images. A homography matrix value is calculated using the coordinates of the feature points of each of the common area and the first image, and the remaining area excluding the area corresponding to the common area is extracted on the first image, thereby extracting the specialized area of ​​the first image. That is, the specialized area of ​​the first image can be extracted using the homography matrix.

[0147] In addition, the first specialized region of the first image extracted from S305 is subjected to homography inverse transformation to extract the first specialized region based on the original pixel coordinates of the first image before coordinate transformation, and pixel coordinates corresponding to the first specialized region can be obtained / stored based on the original pixel coordinates.

[0148] In S305, the transformed second pixel coordinates of the first image are re-transformed into the original second pixel coordinates (S306), and the third pixel coordinates of the first specialized area can be obtained based on the original second pixel coordinates / re-transformed second pixel coordinates of the first image (S307).

[0149] S307 The specialized region of the first image extracted based on the pixel coordinates of the original first image is extracted based on the pixel coordinates of the original first image, so that the original image of the first image can be restored based on the specialized region of the first image and the common region between the plurality of images included in the cluster including the first image.

[0150] By repeatedly performing the implementations of S301 to S307 for each of the plurality of images included in the first cluster, the specialized region of each of the final plurality of images can be extracted.

[0151] Next, with reference to FIG. 10, a cluster generated according to some embodiments of the present disclosure, a common area between multiple images included in the cluster, and a specialized area of ​​each of the multiple images are described.

[0152] Specifically, FIG. 10 is a diagram illustrating an example of a common area of ​​images included in a cluster generated according to some embodiments described with reference to FIGS. 2 to 9 and a specialized area of ​​each image.

[0153] Referring to FIG. 10, according to some embodiments of the present disclosure, a first image (11), a second image (12), and a third image (13) may be clustered into one cluster. In this case, an image area corresponding to an overlapping area at the same location on the first image (11), the second image (12), and the third image (13) may be extracted as a common area (20). In addition, an area excluding the common area (20) on the first image (11) may be extracted as a specialized area (21) of the first image (11), an area excluding the common area (20) on the second image (12) may be extracted as a specialized area (22) of the second image (12), and an area excluding the common area (20) on the third image (13) may be extracted as a specialized area (23) of the third image (13).

[0154] In the present disclosure, specialized regions of multiple images included in one cluster, such as 11, 22, and 23 shown in FIG. 10, can be named a specialized region set.

[0155] Next, an example of the entire operation process of the image storage system (300) is described.

[0156] FIG. 11 is a flowchart illustrating an example of an overall operation process that an image storage system (300) can perform according to some embodiments described with reference to FIGS. 2 to 9.

[0157] Referring to FIG. 11, first, an image storage system (300) that has acquired N (N is a natural number) original images (Image 1 to Image M) can cluster the original images into clusters to generate M (M is a natural number) clusters (Cluster 1 to Cluster M) (S2). Here, M may be equal to or less than N. In addition, S2 may correspond to S100 of FIG. 2.

[0158] And, the image storage system (300) can perform an image flattening operation on each of clusters 1 to M to compress the images included in each of clusters 1 to M (S3).

[0159] And, the image storage system (300) can extract common areas of images included in each of clusters 1 to M (S4). Since common areas are extracted for each cluster, a total of M common areas can be extracted. That is, a total of M image areas can be extracted from S4. S4 can correspond to S200 of FIG. 2. In addition, the common areas extracted from S4 can be stored separately from the specialized areas of each image (S5).

[0160] And, the image storage system (300) can extract and store the specialized regions of each image included in each of clusters 1 to M (S6). Since the specialized regions are extracted for each cluster, a total of M specialized region sets (specialized region set 1 to specialized region set M) can be extracted, and each of the M specialized region sets can include specialized regions of images included in each cluster. For example, specialized region set 1 can include specialized regions of images included in cluster 1, and specialized region set M can include specialized regions of images included in cluster M. S6 can correspond to S300 of FIG. 2.

[0161] Additionally, M original images can be restored based on the set of M extracted common regions and M extracted specialized regions (S7).

[0162] For example, in S7, the original images can be restored by performing homography inverse transformation on the common areas between images included in a cluster and the specialized areas of each image.

[0163] In other words, in S7, the common area can be projected onto each image, and the specific areas of each image can be combined to restore the original images. At this time, according to some embodiments of the present disclosure, images flattened using AVIF can be obtained as restored results.

[0164] Additionally, the restoration process in S7 can be performed cluster by cluster, resulting in restoration of original images clustered into M clusters.

[0165] Although FIG. 11 illustrates the image storage system (300) restoring original images, this is not a limitation. For example, as described with reference to FIG. 1, a server different from the image storage system (300) may receive a set of common and specialized regions extracted by cluster from the image storage system (300) and restore the original images.

[0166] According to some embodiments of the present disclosure described with reference to FIGS. 2 to 11, as the amount of images to be stored / transmitted / received increases, the compression ratio can increase, and as the amount of images included in one cluster increases, a low loss rate and a high compression ratio can be ensured.

[0167] By clustering images into clusters, extracting common areas of the images and specialized areas of each image for each cluster and storing them separately, storage space can be used more efficiently than by slicing the images to separate them into cross-sectional images, and finding and storing changed areas between cross-sectional images of adjacent layers separately. In addition, in the process of restoring images, if the images are sliced ​​and separated into layers, all cross-sectional images must be restored sequentially, whereas by storing them separately into common areas and specialized areas according to some embodiments of the present disclosure, only the images that are specific targets to be restored can be restored, which is efficient.

[0168] FIG. 12 is a block diagram illustrating an example of a computing device for performing some embodiments of the present disclosure.

[0169] Referring to FIG. 12, a computing device (1) may include one or more processors (101), a bus (103), a communication interface (104), a memory (102) for loading a computer program executed by the processor (101), and a storage (105) for storing a computer program (106). However, only components related to the embodiment of the present disclosure are illustrated in FIG. 12. Therefore, a person skilled in the art to which the present disclosure pertains will appreciate that other general components may be included in addition to the components illustrated in FIG. 12. That is, the computing device (1) may further include various components in addition to the components illustrated in FIG. 12. In addition, in some cases, the computing device (1) may be configured in a form in which some of the components illustrated in FIG. 12 are omitted. Hereinafter, each component of the computing device (1) will be described.

[0170] The processor (101) can control the overall operation of each component of the computing device (1). The processor (101) can be configured to include at least one of a CPU (Central Processing Unit), an MPU (Micro Processor Unit), an MCU (Micro Controller Unit), a GPU (Graphics Processing Unit), or any other type of processor well known in the technical field of the present disclosure. In addition, the processor (101) can perform operations for at least one application or program for executing a method and / or operation according to embodiments of the present disclosure. The computing device (1) can be equipped with one or more processors.

[0171] In addition, the system and / or computing device (1) according to embodiments of the present disclosure may further include a database, and the processor (101) may store data and / or information generated / output according to some embodiments of the present disclosure in the memory (102) and / or the database. Here, the database in which the data and / or information is stored is not limited to a database within the system and / or computing device (1) according to the present disclosure, and may include, for example, a database of an external server.

[0172] The memory (102) can store various data, commands, and / or information. The memory (102) can load a computer program (106) from the storage (105) to execute methods and / or operations according to embodiments of the present disclosure. The memory (102) may be implemented as a volatile memory such as RAM, but the present disclosure is not limited thereto.

[0173] The bus (103) can provide a communication function between components of the computing device (1). The bus (103) can be implemented as various types of buses such as an address bus, a data bus, and a control bus.

[0174] The communication interface (104) can support wired / wireless Internet communication of the computing device (1). Furthermore, the communication interface (104) can also support various communication methods other than Internet communication. To this end, the communication interface (104) can be configured to include a communication module well known in the technical field of the present disclosure.

[0175] Storage (105) can non-temporarily store one or more computer programs (106). Storage (105) can be configured to include non-volatile memory such as Read Only Memory (ROM), Erasable Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), flash memory, a hard disk, a removable disk, or any form of computer-readable recording medium well known in the art to which the present disclosure pertains.

[0176] The computer program (106) may include one or more instructions that, when loaded into the memory (102), cause the processor (101) to perform methods and / or operations according to various embodiments of the present disclosure. That is, the processor (101) may perform the methods / implementations / operations according to various embodiments of the present disclosure by executing the one or more loaded instructions.

[0177] In the present disclosure, a computer-readable (non-volatile) storage medium can store at least one instruction or computer program, which when executed by at least one processor can cause the at least one processor to perform a method and / or operation according to some embodiments of the present disclosure.

[0178] For example, the computer program (106) may include instructions for clustering images into clusters, extracting a common area between a plurality of images included in a first cluster included in the clusters, extracting a specialized area of ​​each of the plurality of images included in the first cluster based on the extracted common area, and storing the common area and the specialized area of ​​each of the plurality of images included in the first cluster.

[0179] According to some embodiments of the present disclosure, by extracting and storing common areas between images and specialized areas of each image, the efficiency of utilizing limited storage resources can be increased compared to storing the original images themselves. Furthermore, by transmitting and receiving the common areas between images and specialized areas of each image during the process of transmitting and receiving the images, network traffic can be reduced compared to transmitting and receiving the original images themselves.

[0180] In addition, according to some embodiments of the present disclosure, images are clustered into clusters based on predetermined criteria (e.g., anomaly detection location on the image, code number of an object included in the image, design change reflection time / time interval, etc.), so that a common area can be extracted for each cluster, thereby increasing the image compression ratio compared to a case where a common area of ​​all non-clustered images is extracted.

[0181] Various embodiments of the present disclosure and effects according to the embodiments have been described with reference to FIGS. 1 through 12. The effects according to the technical concept of the present disclosure are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the contents of the present disclosure.

[0182] Although the components constituting the embodiments of the present disclosure have been described as being combined or operating in combination as one, the present disclosure is not necessarily limited to such embodiments. That is, within the scope of the present disclosure, the components may be selectively combined and operated in one or more combinations.

[0183] Although operations are depicted in the drawings in a specific order, this should not be construed as requiring that the operations be performed in the specific order depicted, or in a sequential order, or that all depicted operations be performed to achieve the desired results. In certain circumstances, multitasking and parallel processing may be advantageous. Furthermore, the separation of various components in the embodiments described above should not be construed as requiring such separation, and it should be understood that the program components and systems described may generally be integrated together into a single software product or packaged into multiple software products.

[0184] Although the embodiments of the present disclosure have been described with reference to the attached drawings, those skilled in the art will appreciate that the present disclosure can be implemented in other specific forms without altering the technical spirit or essential features thereof. Therefore, it should be understood that the embodiments described above are exemplary in all respects and not restrictive. The scope of protection of the present disclosure should be interpreted by the claims below, and all technical ideas within a scope equivalent thereto should be interpreted as being included within the scope of the rights of the present disclosure.

Claims

1. In a method performed by a computing system, A step of clustering images into clusters; A step of extracting a common area between a plurality of images included in a first cluster included in the above clusters; A step of extracting a specialized area of ​​each of the plurality of images included in the first cluster based on the extracted common area; and Comprising a step of storing the common area and the specialized area of ​​each of the plurality of images included in the first cluster, How to save images.

2. In paragraph 1, The above images are of each of the multiple items manufactured during the first time period, and The above first time interval includes the time point at which design changes related to the manufacturing of the plurality of items are reflected, The steps to cluster the above images are: A step of dividing the first time interval into a second time interval and a third time interval based on the time point at which the design change is reflected; and A step of clustering images captured in the second time interval into a second cluster, and clustering images captured in the third time interval into a third cluster, How to save images.

3. In paragraph 1, The steps to cluster the above images are: A step of calculating the similarity between the images using SSIM (Structed Similarity Image Matching); and A step of clustering the first image and the second image into one cluster when the similarity between the first image and the second image included in the above images is greater than or equal to a preset threshold value, How to save images.

4. In paragraph 3, The step of extracting the common area between the plurality of images included in the first cluster comprises: A step of calculating the shortest path by applying a shortest path algorithm to the plurality of images based on the calculated similarity between the images; and A step of extracting a common area between images included in the calculated shortest path among the plurality of images, How to save images.

5. In paragraph 1, The plurality of images included in the first cluster include a first image and a second image, The step of extracting the common area between the plurality of images included in the first cluster comprises: A step of extracting first feature points of the first image and second feature points of the second image; A step of generating feature point pairs by matching each of the first feature points with each of the second feature points based on the similarity between each of the first feature points and each of the second feature points; and A step of extracting a partial common area between the first image and the second image based on the feature point pairs, How to save images.

6. In paragraph 5, The step of extracting the partial common area between the first image and the second image based on the feature point pairs is: A step of obtaining first pixel coordinates of the first image and second pixel coordinates of the second image; A step of transforming the second pixel coordinates of the second image so that the pixel coordinates of each of the feature points of the first image and the feature points of the second image included in each of the feature point pairs are identical; and A step of extracting the partial common area by projecting the second image whose coordinates have been transformed onto the first image, How to save images.

7. In paragraph 1, Based on the extracted common area, the step of extracting the specialized area of ​​each of the plurality of images included in the first cluster is: A step of extracting first feature points of the extracted common area and second feature points of the first image included in the plurality of images; A step of generating feature point pairs by matching each of the first feature points with each of the second feature points based on the similarity between each of the first feature points and each of the second feature points; A step of obtaining first pixel coordinates of the extracted common area and second pixel coordinates of the first image; A step of transforming the second pixel coordinates of the first image so that the pixel coordinates of the extracted common area feature points included in each of the feature point pairs and each of the feature points of the first image are identical; and A step of extracting a first specialized area excluding the extracted common area of ​​the coordinate-transformed first image by projecting the coordinate-transformed first image onto the extracted common area, How to save images.

8. In paragraph 7, Based on the extracted common area, the step of extracting the specialized area of ​​each of the plurality of images included in the first cluster is: A step of reconverting the transformed second pixel coordinates of the transformed first image into the second pixel coordinates; and Further comprising a step of obtaining third pixel coordinates of the first specialized area based on the second pixel coordinates of the first image. How to save images.

9. In paragraph 1, For each of the above clusters, further comprising a step of reducing the resolution of a plurality of images included in each cluster. How to save images.

10. In paragraph 5, The first feature points and the second feature points are each extracted based on SIFT (Scale-Invariant-Feature Transform). How to save images.

11. At least one processor; and At least one memory storing instructions that, when executed by at least one processor, cause the at least one processor to perform operations, The above actions are, The act of clustering images into clusters; An operation of extracting a common area between a plurality of images included in a first cluster included in the above clusters; An operation of extracting a specialized area of ​​each of a plurality of images included in the first cluster based on the extracted common area; and Including an operation of storing the specialized area of ​​each of the plurality of images included in the common area and the first cluster, Image storage system.

12. In paragraph 11, The above images are of each of the multiple items manufactured during the first time period, and The above first time interval includes the time point at which design changes related to the manufacturing of the plurality of items are reflected, The operation of clustering the above images is: An operation of dividing the first time interval into a second time interval and a third time interval based on the time point at which the design change is reflected; and An operation of clustering images captured in the second time interval into a second cluster, and clustering images captured in the third time interval into a third cluster, Image storage system.

13. In paragraph 11, The operation of clustering the above images is: An operation of calculating the similarity between the images using SSIM (Structed Similarity Image Matching); and An operation of clustering the first image and the second image into one cluster when the similarity between the first image and the second image included in the above images is greater than or equal to a preset threshold value, Image storage system.

14. In paragraph 13, The operation of extracting the common area between the plurality of images included in the first cluster is: An operation of calculating a shortest path by applying a shortest path algorithm to the plurality of images based on the calculated similarity between the images; and Including an operation of extracting a common area between images included in the calculated shortest path among the plurality of images, Image storage system.

15. In paragraph 11, The plurality of images included in the first cluster include a first image and a second image, The operation of extracting the common area between the plurality of images included in the first cluster is: An operation of extracting first feature points of the first image and second feature points of the second image; An operation of generating feature point pairs by matching each of the first feature points with each of the second feature points based on the similarity between each of the first feature points and each of the second feature points; and Including an operation of extracting a partial common area between the first image and the second image based on the feature point pairs. Image storage system.

16. In paragraph 11, Based on the common area extracted above, the operation of extracting the specialized area of ​​each of the plurality of images included in the first cluster is: An operation of extracting first feature points of the extracted common area and second feature points of the first image included in the plurality of images; An operation of generating feature point pairs by matching each of the first feature points with each of the second feature points based on the similarity between each of the first feature points and each of the second feature points; An operation of obtaining first pixel coordinates of the extracted common area and second pixel coordinates of the first image; An operation of transforming the second pixel coordinates of the first image so that the pixel coordinates of the extracted common area feature points included in each of the feature point pairs and each of the feature points of the first image are identical; and An operation of projecting the coordinate-transformed first image onto the extracted common area, and extracting a first specialized area of ​​the coordinate-transformed first image excluding the extracted common area, Image storage system.

17. In combination with a computing device, A step of clustering images into clusters; A step of extracting a common area between a plurality of images included in a first cluster included in the above clusters; A step of extracting a specialized area of ​​each of the plurality of images included in the first cluster based on the extracted common area; and Stored in a recording medium to execute a step of storing the specialized area of ​​each of the plurality of images included in the common area and the first cluster, Computer program.

Images