Satellite remote sensing image building change detection method
CN116664470BActive Publication Date: 2026-05-26Chinese People's Liberation Army Cyberspace Force Information Engineering University
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Chinese People's Liberation Army Cyberspace Force Information Engineering University
- Filing Date
- 2022-02-18
- Publication Date
- 2026-05-26
AI Technical Summary
Technical Problem
The existing technology suffers from low efficiency due to the step-by-step process of building extraction and transformation detection.
Method used
A building change detection model is adopted, which extracts multi-scale features through the backbone network and uses the building extraction network and change detection network in the head network to simultaneously realize building extraction and change detection. Feature processing is carried out by combining multilayer perceptron and convolutional layers, and a loss function is constructed for constraint.
Benefits of technology
It improves the efficiency of building change detection, simplifies processing steps, and enhances the accuracy of building extraction and change detection.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN116664470B_ABST
Abstract
This invention proposes a method for detecting building changes in satellite remote sensing imagery, belonging to the field of satellite imagery change detection technology. This invention utilizes a pre-trained building change detection model to extract buildings and detect whether changes have occurred. The model includes a backbone network and a head network. First, the backbone network extracts multi-scale features from two input temporal remote sensing images. Then, two building extraction networks within the head network extract buildings from the two temporal remote sensing images respectively. Finally, a change detection network within the head network detects whether changes have occurred in the buildings between the two temporal remote sensing images. Compared to existing technologies that separate building extraction and change detection in remote sensing images, this invention simultaneously achieves building extraction and change detection using this model, improving work efficiency and simplifying processing steps.
Need to check novelty before this filing date? Find Prior Art