Change detection method and system based on multi-scale fusion and integrated kernel adaptive learning

By constructing a Siam sensitivity-enhanced feature extraction network and a multi-scale dual-time fusion attention module, combined with an integrated kernel adaptive learning module, the problems of high computational complexity and feature redundancy in high-resolution image processing are solved, thereby improving the accuracy of change detection and the ability to recognize subtle changes.

CN119107561BActive Publication Date: 2026-06-12NANJING UNIV OF SCI & TECH
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
NANJING UNIV OF SCI & TECH
Filing Date
2024-08-30
Publication Date
2026-06-12

AI Technical Summary

Technical Problem

Existing change detection methods suffer from high computational complexity in high-resolution image processing, low recognition accuracy due to redundant feature extraction, and neglect of important features across two time phases, resulting in low detection accuracy. Furthermore, the complementary relationship between deep and shallow features is not effectively integrated.

Method used

A Siam sensitivity-enhanced feature extraction network is constructed. Through a multi-scale dual-time fusion attention module and an integrated kernel adaptive learning module, feature redundancy is reduced, subtle changes in images are captured, and multi-scale semantic information is fused to enhance feature representation capabilities.

Benefits of technology

It effectively reduces computational complexity, improves the accuracy of change detection, reduces the false negative rate, and enhances the ability to recognize subtle changes in images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119107561B_ABST
    Figure CN119107561B_ABST
Patent Text Reader

Abstract

The application discloses a change detection method and system based on multi-scale fusion and integrated kernel adaptive learning, and the method comprises the following steps: constructing a Thailand sensitivity enhancement feature extraction network for extracting double-time state features; establishing a multi-scale double-time fusion attention module to fuse double-time images of each branch; defining a nested decoder composed of a jump connection mode to retain high-dimensional semantic details and fine-grained positioning details; constructing an integrated kernel adaptive learning module to fuse output features of four levels and fuse multi-scale semantic information; extracting an optimization result and outputting a change area. The application can well alleviate the false change phenomenon caused by feature redundancy in the feature extraction process, more effectively identify important components in double-time features, enhance feature expression diversity, capture subtle changes in images and reduce the missed detection rate in small change areas.
Need to check novelty before this filing date? Find Prior Art

Citation Information

Patent Citations

  • Heterogenous remote sensing image change detection method

    CN117292261A

  • Multi-scale feature interaction network implementation method for processing dual-temporal remote sensing image change

    CN117876782A