Change Detection Based on Artificial Intelligence: State-of-the-Art and Challenges

Abstract

Change detection based on remote sensing (RS) data is an important method of detecting changes on the Earth’s surface and has a wide range of applications in urban planning, environmental monitoring, agriculture investigation, disaster assessment, and map revision. In recent years, integrated artificial intelligence (AI) technology has become a research focus in developing new change detection methods. Although some researchers claim that AI-based change detection approaches outperform traditional change detection approaches, it is not immediately obvious how and to what extent AI can improve the performance of change detection. This review focuses on the state-of-the-art methods, applications, and challenges of AI for change detection. Specifically, the implementation process of AI-based change detection is first introduced. Then, the data from different sensors used for change detection, including optical RS data, synthetic aperture radar (SAR) data, street view images, and combined heterogeneous data, are presented, and the available open datasets are also listed. The general frameworks of AI-based change detection methods are reviewed and analyzed systematically, and the unsupervised schemes used in AI-based change detection are further analyzed. Subsequently, the commonly used networks in AI for change detection are described. From a practical point of view, the application domains of AI-based change detection methods are classified based on their applicability. Finally, the major challenges and prospects of AI for change detection are discussed and delineated, including (a) heterogeneous big data processing, (b) unsupervised AI, and (c) the reliability of AI. This review will be beneficial for researchers in understanding this field.

A Novel Change Detection Method for Natural Disaster Detection and Segmentation from Video Sequence

“…However, these traditional optical flow computation methods often provide blurred boundaries and are hard to be used in real time [ 46 , 47 ]. Convolutional neural networks (CNNs) have a strong ability of feature extraction and speckle noise suppressing [ 15 , 48 , 49 ], which has attracted more attention to numerous computer vision tasks.…”

“…With the continuing growth of earth observation techniques and computer technology, massive amounts of remote sensing data for natural disaster with different spectral-spatial-temporal resolution are available for surveying and assessing changes in natural disaster, which greatly promotes the development of change detection methodologies. Many change detection approaches for natural disaster scenes have been proposed and they can be broadly divided into traditional and deep learning (DL)-based [ 15 ]. For traditional CD methods, the simplest approaches are algebra-based methods.…”

Unsupervised Change Detection Using Multiscale and Multiresolution Gaussian-Mixture-Model Guided by Saliency Enhancement

“…Currently, a lot of change detection methods have been reported to detect the changed information on this earth we live. These change detection methods can be roughly grouped into three categories: pixel-based approaches [8][9][10][11][12][13][14][15][16][17][18][19], objectbased approaches [20][21][22][23][24][25], and deep learning (DL) based approaches [26][27][28][29][30][31][32][33][34].…”

“…Driven by the complex image data and wide application of change detection, many DL-based approaches [26][27][28][29][30][31][32][33][34] have been widely used for change detection tasks since they show powerful ability on feature mining. Generally, change detection concerns three types of images, i.e., Synthetic Aperture Radar (SAR) images, VHR images, and hyperspectral images.…”

“…More recently, Peng et al [30] presented a comprehensive summary of deep learning change detection (DLCD) techniques and proposed an improved UNet++ for change detection tasks to capture subtle changes in challenging scenes. In addition, in order to summarize the development and challenges of change detection, Shi et al [32] presented a survey that showed there was an upward trend in the current publicly change detection datasets, but only a few of them were labeled due to the difficulty of annotation. To sum up, the DL-based approaches provide excellent change detection results due to the strong ability of feature representation.…”