US-12626479-B2 - Difference detection device and method

US12626479B2US 12626479 B2US12626479 B2US 12626479B2US-12626479-B2

Abstract

A difference detection device comprising: a data collection module configured to collect a satellite image set comprising a plurality of satellite images; a pre-processing module configured to generate a pre-processing image set by matching characteristic information of the satellite images included in the satellite image set; and a difference determination module configured to generate difference information between the plurality of satellite images by determining a difference between pre-processing images included in the pre-processing image set, wherein the characteristic information comprises at least one of illumination information and color information of the satellite image.

Inventors

Chan hee CHOI
HYUN SUN PARK
Dong Young Kim

Assignees

MEISSA PLANET INC.

Dates

Publication Date: 20260512
Application Date: 20231222
Priority Date: 20230913

Claims (6)

1 . A difference detection device comprising: a processor; a memory operatively coupled to the processor and storing instructions that, when executed by the processor, cause the processor to implement: a data collection module configured to collect a satellite image set comprising a plurality of satellite images; a pre-processing module configured to generate a pre-processing image set by matching characteristic information of the satellite images included in the satellite image set; and a difference determination module configured to generate difference information between the plurality of satellite images by determining a difference between pre-processing images included in the pre-processing image set, wherein the characteristic information comprises at least one of illumination information and color information of the satellite image, wherein the pre-processing module generates the pre-processing image set based on a predefined reference image, wherein the pre-processing module determines characteristic information of the reference image as reference characteristic information by extracting the characteristic information of the reference image, and changes the characteristic information of each of the satellite images included in the satellite image set based on the determined reference characteristic information.
2 . The difference detection device of claim 1 , wherein: the satellite image set comprises a first satellite image photographed at first timing and a second satellite image photographed at second timing different from the first timing, and the first satellite image and the second satellite image comprise photographed satellite images of an identical coordinate region.
3 . The difference detection device of claim 1 , wherein the pre-processing module changes the characteristic information of each of the satellite images included in the satellite image set by using a predefined image processing algorithm.
4 . The difference detection device of claim 3 , wherein the image processing algorithm comprises at least one of image histogram matching and feature distribution matching.
5 . A difference detection device comprising: a processor; a memory operatively coupled to the processor and storing instructions that, when executed by the processor, cause the processor to implement: a data collection module configured to collect a satellite image set comprising a plurality of satellite images; a pre-processing module configured to generate a pre-processing image set by matching characteristic information of the satellite images included in the satellite image set; and a difference determination module configured to generate difference information between the plurality of satellite images by determining a difference between pre-processing images included in the pre-processing image set, wherein the characteristic information comprises at least one of illumination information and color information of the satellite image, wherein the difference determination module generates the difference information by using a deep learning technology, wherein the difference determination module generates the difference information based on a predefined difference detection algorithm in relation to difference detection between a plurality of images, wherein the difference detection device further comprising a learning module configured to train an adaptive difference detection algorithm in which an attention module has been added to the difference detection algorithm, wherein the difference determination module generates the difference information by using the adaptive difference detection algorithm.
6 . The difference detection device of claim 5 , wherein the adaptive difference detection algorithm comprises: a first attention module that is an attention module relating to a channel exchange module for changing at least one of pieces of RGB information in each of the pre-processing images of the pre-processing image set, and a second attention module that is an attention module relating to a spatial exchange module for exchanging at least one pixel of each of the pre-processing images of the pre-processing image set at specific coordinates.

Description

GOVERNMENT SUPPORT CLAUSE This invention has been supported by the Ministry of Trade, Industry and Energy, and the Defense Acquisition Program Administration (Agency for Defense Development Cooperation). The notice number or announcement number of the research project is 2022-DC-2, and the name of the research project is “Support for Commercialization of Defense Technologies”. The research project number is 22-DC-IN-17, and the title of the research project is “Feasibility Study of ‘Aerial Imagery-based AI Smart Construction Service’ commercialization”. The research period is from Dec. 1, 2022, to Nov. 30, 2024. BACKGROUND CROSS-REFERENCE TO RELATED APPLICATION This application claims priority under 35 U.S.C § 119 to Korean Patent Application No. 10-2023-0122030 filed on Sep. 13, 2023, in the Korean Intellectual Property Office, the entire contents of which is hereby incorporated by reference. BACKGROUND 1. Technical Field The present disclosure relates to a difference detection device and method. Particularly, the present disclosure relates to a difference detection device and method capable of improving robustness in the detection of a change in a building by matching characteristic information (e.g., distributions) between multi-time satellite images. Furthermore, the present disclosure relates to a difference detection device and method capable of improving robustness and performance for difference detection by using an adaptive difference detection algorithm in which an attention module has been added to a difference detection algorithm. 2. Related Art Contents described in this part merely provide background information of the present embodiment and do not constitute a conventional technology. Currently, several hundreds of satellites perform their unique tasks above the earth for weather, telecommunications, broadcasting, agriculture, space development, and military purposes. Satellite images that are photographed by the satellites are public data and are variously used in several fields. The satellite image is an image that is photographed at a very high altitude. Accordingly, a technology for processing, analyzing and/or storing the satellite image by using a deep learning technology is recently in the spotlight. In this case, a difference detection technology, that is, one of utilization points for the analysis of a satellite image, is a technology for checking the new construction or extinction of a building within a corresponding time interval by comparing and analyzing multi-time satellite images. However, in a current difference detection technology, improving the robustness of a difference detection model is merely attempted through data augmentation, but attempts to perform accurate difference detection by matching characteristic information (e.g., distributions) between multi-time satellite images are not present. Accordingly, there is a sufficient need for a technology capable of increasing the robustness of difference detection through pre-processing by matching characteristic information (e.g., distributions) between multi-time satellite images. The description set forth in the background section should not be assumed to be prior art merely because it is set forth in the background section. The background section may describe aspects or embodiments of the disclosure. SUMMARY Various embodiments are directed to providing a difference detection device and method capable of improving robustness in the detection of a change in a building, for example, by matching characteristic information between multi-time satellite images. Specifically, various embodiments are directed to providing a difference detection device and method capable of detecting a change in a building more accurately by matching pieces of illumination information and pieces of color information between multi-time satellite images by using a reference image. Also, various embodiments are directed to providing a difference detection device and method capable of improving the robustness of difference detection and performance by using an adaptive difference detection algorithm in which an attention module has been added to a difference detection algorithm. Specifically, various embodiments are directed to providing a difference detection device and method capable of performing an exchange process using each channel and the importance of a space by additionally using an attention module relating to a channel exchange module for exchanging any one of pieces of RGB information in each of images and an attention module relating to a spatial exchange module for exchanging pixels of the images at specific coordinates, and thus improving the robustness and performance of difference detection and. Objects of the present disclosure are not limited to the aforementioned objects, and other objects and advantages of the present disclosure not described above may be understood from the following description and more evidently understood bas