A method, system and storage medium for detecting building changes using vectorized remote sensing images

Abstract

A method for vectorized building change detection using remote sensing images comprises the following steps: 1. constructing a training set and a test set; 2. constructing a vectorized building change detection module, the vectorized building change detection module comprising a feature encoder, a change decoder, a node decoder, and a semantic decoder; 3. inputting the images before and after the change into the feature encoder and subtracting them to obtain a change feature map; then inputting the map into the change decoder to obtain a change detection result; inputting the building image into the feature encoder for feature extraction, and into the node decoder and semantic decoder to obtain node positions and semantic segmentation results; 4. calculating the total loss, repeating steps 3 to 4 until a set number of times is reached; and 5. testing using the test set. The present invention achieves end-to-end vectorized building change detection by jointly training with a building extraction dataset to extract accurate building nodes and change detection results, achieving high vectorization accuracy and concise visual effects.

Images