Unsupervised Change Detection in Satellite Images Using Convolutional\n Neural Networks

2018/12/14 by Kevin Louis de Jong, de Jong, Kevin Louis, Anna Sergeevna Bosman +1 · 1 citation
Earth and Planetary Sciences · Engineering · #FOS: Computer and information sciences #Neural and Evolutionary Computing (cs.NE) #Remote Sensing and Land Use #Remote-Sensing Image Classification

paper · pdf · doi:10.48550/arxiv.1812.05815

openalex publication_date 2018/12/14 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

This paper proposes an efficient unsupervised method for detecting relevant\nchanges between two temporally different images of the same scene. A\nconvolutional neural network (CNN) for semantic segmentation is implemented to\nextract compressed image features, as well as to classify the detected changes\ninto the correct semantic classes. A difference image is created using the\nfeature map information generated by the CNN, without explicitly training on\ntarget difference images. Thus, the proposed change detection method is\nunsupervised, and can be performed using any CNN model pre-trained for semantic\nsegmentation.\n

Cited by

• Efficient Pyramidal Analysis of Gigapixel Images on a Decentralized Modest Computer Cluster

Related

• Unsupervised convolutional neural network fusion approach for change detection in remote sensing images
• Deep learning approaches to Earth Observation change detection
• Fully Convolutional Siamese Networks for Change Detection
• Fully Convolutional Change Detection Framework with Generative Adversarial Network for Unsupervised, Weakly Supervised and Regional Supervised Change Detection
• Urban Change Detection by Fully Convolutional Siamese Concatenate Network with Attention