KR-102962358-B1 - SERVER AND METHOD FOR SUPER-RESOLUTION AND SYSTEM INCLUDING THE SAME

KR102962358B1KR 102962358 B1KR102962358 B1KR 102962358B1KR-102962358-B1

Abstract

The present invention relates to a super-resolution server and method capable of varying the super-resolution method for image data (e.g., satellite imagery, aerial imagery, drone imagery, etc.) according to user requirements, and a system including the same. The super-resolution server may include a data collection module that receives image data and request information related to a user's needs regarding the image data, a super-resolution module that super-resolutions the image data according to the request information to generate a high-resolution image, and an output module that outputs the high-resolution image and reliability information regarding the high-resolution image.

Inventors

박현선
이용주
이여울
사공용협
김동철
구아롬
강명훈

Assignees

주식회사 메이사

Dates

Publication Date: 20260511
Application Date: 20241104

Claims (20)

A data collection module that receives video data and request information related to the user's needs regarding the video data; A super-resolution module that generates a high-resolution image by super-resolution the image data according to the above-mentioned request information; and It includes an output module that outputs the above high-resolution image and reliability information for the above high-resolution image, wherein The above image data includes satellite images taken from a satellite, and The above request information is, Coordinate information related to the coordinate area that the above user intends to analyze, and It includes purpose information related to the reason why the user intends to perform the super-resolution for a location corresponding to the coordinate information and the purpose to be achieved through the high-resolution image, and The above purpose information is, Simple super-resolution meaning only of super-resolution for the above image data and Includes the user's selection of any one of the special analyses, which signifies specific data analysis through the above-mentioned video data, and The above special analysis includes object analysis for analyzing objects included in the satellite image, area analysis for analyzing the area occupied by the objects in the satellite image, and band analysis for analyzing bands for vegetation included in the satellite image. The above selection includes the user performing a click input or touch input for any one of the simple super-resolution, the object analysis, the area analysis, and the band analysis through a user terminal for the user, and The above super-resolution module includes a method determination unit that determines a super-resolution method, which is a method for performing the super-resolution based on the above-mentioned request information. The above method determining unit is, A first method based on up-scaling and One of the second methods based on a super-resolution model using a neural network is determined as the super-resolution method, and The above method determining unit determines the super-resolution method based on the above objective information included in the above request information, If the above-mentioned target information is the above-mentioned simple super-resolution, the above-mentioned second method is determined as the above-mentioned super-resolution method, and If the above target information is the above object analysis among the above special analyses, the above second method is determined as the above super-resolution method, and If the above-mentioned target information is the area analysis or band analysis for a specific target among the above-mentioned special analyses, the result of combining the above-mentioned first method and the above-mentioned second method is determined by the above-mentioned super-resolution method, If the specific object is a road and the type of special analysis is an area analysis intended to analyze the area of the road through the satellite image, the method determining unit determines the area including the road as the region of interest, determines the first method as the super-resolution method for the region of interest including the road, and determines the second method as the super-resolution method for other areas excluding the road. If the specific target is vegetation and the type of special analysis is band analysis intended to analyze bands for the vegetation through the satellite image, the method determining unit determines the area containing the vegetation as the region of interest, determines the first method as the super-resolution method for the region of interest containing the vegetation, and determines the second method as the super-resolution method for other areas excluding the vegetation. The above output module is, Detection of objects included in each of the satellite image and the high-resolution image is performed using a pre-trained detection model, and Generating the degree of agreement between the detection results for the object in the satellite image and the high-resolution image as the reliability information Super-resolution server.
In Article 1, The above data collection module is, Receiving the image data from an external database that stores and manages the image data, and Receiving the request information from the user terminal of the above user Super-resolution server.
delete
delete
In Article 1, The above super-resolution module is, An object segmentation unit that generates a segmentation result for the image data by segmenting an object included in the image data, and The generating unit further includes a generating unit that generates the high-resolution image by performing the super-resolution on the image data according to the super-resolution method determined above. The above method determining unit determines the super-resolution method based on the above division result and the above requirement information. Super-resolution server.
In Article 5, The object segmentation unit generates the segmentation result by segmenting the object using a pre-trained segmentation model based on a neural network. Super-resolution server.
In Article 6, The above segmentation model includes a semantic segmentation model. Super-resolution server.
In Article 5, The above super-resolution model includes a Generative Artificial Intelligence model. Super-resolution server.
delete
delete
delete
delete
In Article 8, The above method determining unit is, For some of the multiple regions included in the above division result, the above first method is determined as the above super-resolution method, and For a part of the plurality of regions that is different from the part mentioned above, the second method is determined as the super-resolution method. Super-resolution server.
In Article 13, The above method determining unit is, Based on the specific target mentioned above, determine the region of interest among the multiple regions included in the division result, and For the determined region of interest above, the first method is determined as the super-resolution method, and For regions other than the aforementioned region of interest, the above second method is determined as the above super-resolution method. Super-resolution server.
delete
delete
In Article 1, The output module above generates the reliability information based on the difference between the image data and the high-resolution image. Super-resolution server.
In Article 17, The output module compares the image data and the high-resolution image to generate the probability of agreement between an object included in the image data and an object included in the high-resolution image as the reliability information. Super-resolution server.
Memory for storing at least one instruction; and It includes at least one processor that executes the above at least one instruction, and The above processor is, Receives video data and request information related to the user's needs regarding the video data, and Based on the above requirement information, the above image data is super-resolved to generate a high-resolution image, and Output the above high-resolution image and reliability information for the above high-resolution image, The above image data includes satellite images taken from a satellite, and The above request information is, Coordinate information related to the coordinate area that the above user intends to analyze, and It includes purpose information related to the reason why the user intends to perform the super-resolution for a location corresponding to the coordinate information and the purpose to be achieved through the high-resolution image, and The above purpose information is, Simple super-resolution meaning only of super-resolution for the above image data and Includes the user's selection of any one of the special analyses, which signifies specific data analysis through the above-mentioned video data, and The above special analysis includes object analysis for analyzing objects included in the satellite image, area analysis for analyzing the area occupied by the objects in the satellite image, and band analysis for analyzing bands for vegetation included in the satellite image. The above selection includes the user performing a click input or touch input for any one of the simple super-resolution, the object analysis, the area analysis, and the band analysis through a user terminal for the user, and The above processor is, Based on the above-mentioned requirement information, determine a super-resolution method that is a method for performing the above-mentioned super-resolution, A first method based on up-scaling and One of the second methods based on a super-resolution model using a neural network is determined as the super-resolution method, The above processor determines the super-resolution method based on the above-mentioned objective information included in the above-mentioned request information, If the above-mentioned target information is the above-mentioned simple super-resolution, the above-mentioned second method is determined as the above-mentioned super-resolution method, and If the above target information is the above object analysis among the above special analyses, the above second method is determined as the above super-resolution method, and If the above-mentioned target information is the area analysis or band analysis for a specific target among the above-mentioned special analyses, the result of combining the above-mentioned first method and the above-mentioned second method is determined by the above-mentioned super-resolution method, If the above specific object is a Road and the type of special analysis is an area analysis intended to analyze the area of the Road through the satellite image, the processor determines the area containing the Road as a region of interest, determines the first method as the super-resolution method for the region of interest containing the Road, and determines the second method as the super-resolution method for other areas excluding the Road. If the specific target is vegetation and the type of special analysis is band analysis intended to analyze bands for the vegetation through the satellite image, the processor determines the area containing the vegetation as a region of interest, determines the first method as the super-resolution method for the region of interest containing the vegetation, and determines the second method as the super-resolution method for other areas excluding the vegetation. The above processor is, Detection of objects included in each of the satellite image and the high-resolution image is performed using a pre-trained detection model, and Generating the degree of agreement between the detection results for the object in the satellite image and the high-resolution image as the reliability information Super-resolution server.
In Article 19, The above processor is, Receiving the image data from an external database that stores and manages the image data, and Receiving the request information from the user terminal of the above user Super-resolution server.

Description

Super-resolution server and method and system including the same The present invention relates to a super-resolution server, a method, and a system including the same. Specifically, the present invention relates to a super-resolution server and method capable of varying the super-resolution method for image data (e.g., satellite imagery, aerial imagery, drone imagery, etc.) according to user requirements, and a system including the same. The content described in this section merely provides background information regarding the present embodiment and does not constitute prior art. Recently, image data, such as satellite imagery, aerial imagery, and drone footage, is being utilized to acquire information about buildings, roads, and nature. This image data is being used in various fields, including detecting the risk of wildfires and assessing vegetation indices. In this context, particularly since satellite imagery is captured from the ground up to high altitudes, there are cases where data analysis is difficult due to low resolution when utilizing such images. In such cases, super-resolution is performed to increase the resolution of the satellite imagery. However, generally, when performing super-resolution on satellite imagery, only known post-processing methods are executed in a batch manner, and there is no technological development regarding super-resolution methods tailored to specific user needs. Accordingly, there is a sufficient need for super-resolution related technologies that reflect user requirements. FIG. 1 illustrates a super-resolution system according to some embodiments of the present invention. FIG. 2 is a diagram illustrating the neural network structure of a neural network model used by a super-resolution server according to some embodiments of the present invention. FIG. 3 is a block diagram of a super-resolution server according to some embodiments of the present invention. Figure 4 illustrates image data according to some embodiments of the present invention. FIGS. 5 to 7 are drawings for explaining requirement information according to some embodiments of the present invention. FIG. 8 is a block diagram of a super-resolution module according to some embodiments of the present invention. FIG. 9 is a detailed block diagram of a super-resolution module according to some embodiments of the present invention. FIG. 10 is a conceptual diagram illustrating the operation of a super-resolution module according to some embodiments of the present invention. FIG. 11 is a detailed block diagram of an object segmentation unit included in a super-resolution module according to some embodiments of the present invention. FIGS. 12 and 13 are drawings for explaining the learning and execution steps of a segmentation model used by an object segmentation unit according to some embodiments of the present invention. FIG. 14 is a diagram illustrating a super-resolution method according to some embodiments of the present invention. FIG. 15 is a diagram illustrating the operation of a method determination unit in the case where the target information according to some embodiments of the present invention is simple super-resolution. FIG. 16 is a diagram illustrating the operation of a method determination unit in the case where the objective information according to some embodiments of the present invention is object analysis. FIG. 17 is a diagram illustrating the operation of a method determination unit in the case where the objective information according to some embodiments of the present invention is area analysis or band analysis. FIG. 18 is a conceptual diagram for explaining the operation of a method determination unit in the case where the objective information according to some embodiments of the present invention is an analysis of the area of a road. FIG. 19 is a conceptual diagram illustrating the operation of a method determination unit in the case where the objective information according to some embodiments of the present invention is a band analysis of vegetation. FIG. 20 is a block diagram of an output module according to some embodiments of the present invention. FIG. 21 is a drawing for explaining reliability information according to some embodiments of the present invention. FIG. 22 is a block diagram of a super-resolution server according to some other embodiments of the present invention. FIG. 23 is a flowchart of a super-resolution method according to some embodiments of the present invention. FIG. 24 is a detailed flowchart of step (S100) of FIG. 23 according to some embodiments of the present invention. FIG. 25 is a detailed flowchart of step (S200) of FIG. 23 according to some embodiments of the present invention. FIG. 26 is a detailed flowchart of step (S210) of FIG. 25 according to some embodiments of the present invention. FIG. 27 is a detailed flowchart of step (S220) of FIG. 25 according to some embodiments of the present invention. FIG. 28 is a detailed flowchart of step (S223) of