Change Detection In Satellite Images Research Articles

Hierarchical Feature Association and Global Correction Network for Change Detection

Optical satellite image change detection has attracted extensive research due to its comprehensive application in earth observation. Recently, deep learning (DL)-based methods have become dominant in change detection due to their outstanding performance. Remote sensing (RS) images contain different sizes of ground objects, so the information at different scales is crucial for change detection. However, the existing DL-based methods only employ summation or concatenation to aggregate several layers of features, lacking the semantic association of different layers. On the other hand, the UNet-like backbone is favored by deep learning algorithms, but the gradual downscaling and upscaling operation in the mainstream UNet-like backbone has the problem of misalignment, which further affects the accuracy of change detection. In this paper, we innovatively propose a hierarchical feature association and global correction network (HFA-GCN) for change detection. Specifically, a hierarchical feature association module is meticulously designed to model the correlation relationship among different scale features due to the redundant but complementary information among them. Moreover, a global correction module on Transformer is proposed to alleviate the feature misalignment in the UNet-like backbone, which, through feature reuse, extracts global information to reduce false alarms and missed alarms. Experiments were conducted on several publicly available databases, and the experimental results show the proposed method is superior to the existing state-of-the-art change detection models.

A novel framework for fine-grained spatio-temporal change detection in satellite images

A novel framework for fine-grained spatio-temporal change detection in satellite images

Wader hunt optimization based UNET model for change detection in satellite images

Wader hunt optimization based UNET model for change detection in satellite images

GAN-FuzzyNN: Optimization Based Generative Adversarial Network and Fuzzy Neural Network Classification for Change Detection in Satellite Images

GAN-FuzzyNN: Optimization Based Generative Adversarial Network and Fuzzy Neural Network Classification for Change Detection in Satellite Images

Assessment of hail damages in maize using remote sensing and comparison with an insurance assessment: A case study in Lombardy

Studies have shown that the quantification of hail damage is generally inaccurate and is influenced by the experience of the field surveyors/technicians. To overcome this problem, the vegetation indices retrieved by remote sensing, can be used to get information about the hail damage. The aim of this work is the detection of medium-low damages (i.e., between 10 and 30% of the gross saleable production) using the much-used normalized difference vegetation index (NDVI) in comparison with alternative vegetation indices (i.e., ARVI, MCARI, SAVI, MSAVI, MSAVI2) and their change from pre-event to post-event in five hailstorms in Lombardy in 2018. Seventy-four overlapping scenes (10% cloud cover) were collected from the Sentinel-2 in the spring-summer period of 2018 in the Brescia district (Lombardy). An unsupervised classification was carried out to automatically identify the maize fields (grain and silage), testing the change detection approach by searching for damage by hail and strong wind in the Lombardy plain of Brescia. A database of 125 field surveys (average size 4 Ha) after the hailstorm collected from the insurance service allowed for the selection of the dates on which the event occurred and provided a proxy of the extent of the damage (in % of the decrease of the yield). Hail and strong wind damages ranged from 5 to 70%, and they were used for comparison with the satellite image change detection. The differences in the vegetation indices obtained by Sentinel 2 before and after the hailstorm and the insurance assessments of damage after the events were compared to assess the degree of concordance. The modified soil-adjusted vegetation index outperformed other vegetation indices in detecting hail-related damages with the highest accuracy (73.3%). On the other hand, the NDVI resulted in scarce performance ranking last of the six indices, with an accuracy of 65.3%. Future research will evaluate how much uncertainty can be found in the method’s limitations with vegetation indices derived from satellites, how much is due to errors in estimating damage to the ground, and how much is due to other causes. Highlights - The discovery rate of damaged fields improved. - MSAVI outperformed NDVI and other vegetation indices, identifying 73.3% of occurrences. - Estimation of damage from remote sensing was more accurate for fields severely affected >50%. - In low-intensity hail events (<50 canopies affected), the MSAVI provided a detailed picture of the damage across the field. - The proposed approach is promising to develop a ‘sampling map’ for detailed on-ground assessment.

Cascaded U-Net with Training Wheel Attention Module for Change Detection in Satellite Images

Change detection is an important application of remote sensing image interpretation, which identifies changed areas of interest from a pair of bi-temporal remote sensing images. Various deep-learning-based approaches have demonstrated promising results and most of these models used an encoder–decoder shape such as U-Net for segmentation of changed areas. In order to obtain more refined features, this paper introduces a change detection model with cascaded U-Net. The proposed network architecture contains four cascaded U-Nets with ConvNeXT blocks. With a patch embedding layer, the cascaded structure can improve detection results with acceptable computational overhead. To facilitate the training of the cascaded N-Nets, we proposed a novel attention mechanism called the Training whEel Attention Module (TEAM). During the training phase, TEAM aggregates outputs from different stages of cascaded structures and shifts attention from outputs from shallow stages to outputs from deeper stages. The experimental results show that our cascaded U-Net architecture with TEAM achieves state-of-the-art performance in two change detection datasets without extra training data.

A domain adaptation neural network for change detection with heterogeneous optical and SAR remote sensing images

• A novel pseudo-Siamese network for heterogeneous satellite image change detection is proposed. • A domain adaptation constraint block is proposed to implicitly align the heterogeneous deep features. • Attention-regulated feature refinement improves change map reconstruction. • A large heterogeneous SAR and optical remote sensing change detection dataset is newly created. • Experiments on scenarios of flooding and urban development demonstrate the improved performance of the proposed network. Heterogeneous remote sensing source-based change detection with optical and SAR data and their combined all-time and all-weather observation capability provides a reliable and promising solution for a wide range of applications. State-of-the-art supervised methods typically take a two-stage strategy that suffers from the loss of original image features and the introduction of noise on the transferred images. This paper proposes a domain adaptation-based multi-source change detection network (DA-MSCDNet) suitable to process heterogeneous optical and SAR images. DA-MSCDNet employs feature-level transformation to align inconsistent deep feature spaces in heterogeneous data. Feature space transformation and change detection are bridged within the network to encourage task communication. Experiments are conducted on two public datasets based on Sentinel-1A and Landsat-8 imagery acquired over the Sacramento, Yuba, and Sutter Counties (California, USA), and QuickBird-2 and TerraSAR-X imagery over Gloucester (UK), as well as one new large-scale dataset of Sentinel-2 and COSMO-SkyMed imagery over Wuhan (China). Compared with other six supervised and unsupervised approaches, the proposed method achieves the highest performance with an average precision of 80.81%, recall of 84.39%, mIOU of 73.67% and F1 score of 82.58%, beating the state-of-the-art method with 5.42% improvements on F1 score and 10 times efficiency on training time cost on the large-scale change detection task.

Unsupervised Change Detection in Satellite Images With Generative Adversarial Network

Detecting changed regions in paired satellite images plays a key role in many remote sensing applications. The evolution of recent techniques could provide satellite images with very high spatial resolution (VHR) but made it challenging to apply image coregistration, and many change detection methods are dependent on its accuracy.Two images of the same scene taken at different time or from different angle would introduce unregistered objects and the existence of both unregistered areas and actual changed areas would lower the performance of many change detection algorithms in unsupervised condition.To alleviate the effect of unregistered objects in the paired images, we propose a novel change detection framework utilizing a special neural network architecture -- Generative Adversarial Network (GAN) to generate many better coregistered images. In this paper, we show that GAN model can be trained upon a pair of images through using the proposed expanding strategy to create a training set and optimizing designed objective functions. The optimized GAN model would produce better coregistered images where changes can be easily spotted and then the change map can be presented through a comparison strategy using these generated images explicitly.Compared to other deep learning-based methods, our method is less sensitive to the problem of unregistered images and makes most of the deep learning structure.Experimental results on synthetic images and real data with many different scenes could demonstrate the effectiveness of the proposed approach.

A Novel Method for Segmentation and Change Detection of Satellite Images Using Proximal Splitting Algorithm and Multiclass SVM

This paper presents a novel method for segmentation and change detection of multispectral images using proximal splitting-based clustering and multiclass support vector machine (MSVM). Initially, the multitemporal satellite images are preprocessed and then textures are extracted using Difference of Offset Gaussian filter. In general, the traditional clustering method uses Euclidean distance as a prime factor for segmentation process. For multitextured images such as remotely sensed images, this metric provides inconsistent output. To achieve better segmented results, proximal splitting algorithm has been proposed. This method has been considered as a solution for iterative minimization problem, which is required to find exact changes between the multitemporal images. The MSVM is chosen to group the segmented clusters into a fixed number of classes, since the clusters obtained from the proximal splitting algorithm are not independent with each other. Then, the classified images are subjected to image differencing method to detect the changes. Experimentation is performed with two real data sets of Landsat7 images, which illustrates that the mean of difference in area obtained by the proposed method is reduced by an average of 35.24% compared to the conventional system. The validity index obtained for data set 1 using proposed algorithm is lower than the existing methods.

Comparing experts and novices in Martian surface feature change detection and identification

Change detection in satellite images is a key concern of the Earth Observation field for environmental and climate change monitoring. Satellite images also provide important clues to both the past and present surface conditions of other planets, which cannot be validated on the ground. With the volume of satellite imagery continuing to grow, the inadequacy of computerised solutions to manage and process imagery to the required professional standard is of critical concern. Whilst studies find the crowd sourcing approach suitable for the counting of impact craters in single images, images of higher resolution contain a much wider range of features, and the performance of novices in identifying more complex features and detecting change, remains unknown.This paper presents a first step towards understanding whether novices can identify and annotate changes in different geomorphological features. A website was developed to enable visitors to flick between two images of the same location on Mars taken at different times and classify 1) if a surface feature changed and if so, 2) what feature had changed from a pre-defined list of six. Planetary scientists provided “expert” data against which classifications made by novices could be compared when the project subsequently went public.Whilst no significant difference was found in images identified with surface changes by expert and novices, results exhibited differences in consensus within and between experts and novices when asked to classify the type of change. Experts demonstrated higher levels of agreement in classification of changes as dust devil tracks, slope streaks and impact craters than other features, whilst the consensus of novices was consistent across feature types; furthermore, the level of consensus amongst regardless of feature type. These trends are secondary to the low levels of consensus found, regardless of feature type or classifier expertise. These findings demand the attention of researchers who want to use crowd-sourcing for similar scientific purposes, particularly for the supervised training of computer algorithms, and inform the scope and design of future projects.

Unsupervised change detection of satellite images using low rank matrix completion

Traditional unsupervised change detection methods need to generate a difference image (DI) for subsequent processing to produce a binary change map. In addition, few methods explore global structures. This Letter presents a novel unsupervised change detection approach based on low rank matrix completion. Other than generating a DI, the changed pixels are modeled as the estimated missing values for matrix completion, where the changed pixels are represented by a sparse term. A common low rank matrix is recovered by two temporal images. The changed pixels are separated out from the low rank matrix, in which the local information is introduced via graph cuts. The global and local structures are utilized in our model. Experimental results validate the effectiveness of the proposed approach. The proposed method is a new view for change detection.

Unsupervised change detection in satellite images using fuzzy c-means clustering and principal component analysis

Abstract. Change detection analyze means that according to observations made in different times, the process of defining the change detection occurring in nature or in the state of any objects or the ability of defining the quantity of temporal effects by using multitemporal data sets. There are lots of change detection techniques met in literature. It is possible to group these techniques under two main topics as supervised and unsupervised change detection. In this study, the aim is to define the land cover changes occurring in specific area of Kayseri with unsupervised change detection techniques by using Landsat satellite images belonging to different years which are obtained by the technique of remote sensing. While that process is being made, image differencing method is going to be applied to the images by following the procedure of image enhancement. After that, the method of Principal Component Analysis is going to be applied to the difference image obtained. To determine the areas that have and don’t have changes, the image is grouped as two parts by Fuzzy C-Means Clustering method. For achieving these processes, firstly the process of image to image registration is completed. As a result of this, the images are being referred to each other. After that, gray scale difference image obtained is partitioned into 3 × 3 nonoverlapping blocks. With the method of principal component analysis, eigenvector space is gained and from here, principal components are reached. Finally, feature vector space consisting principal component is partitioned into two clusters using Fuzzy C-Means Clustering and after that change detection process has been done.

Unsupervised Change Detection of Satellite Images Using Local Gradual Descent

In this paper, we propose a novel technique for unsupervised change detection of multitemporal satellite images using Gaussian mixture model (GMM), local gradual descent, and <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$k$</tex> </formula> -means clustering. Data distribution of the difference image is first modeled by bimodal GMM with “changed” and “unchanged” components. The neighborhood data around each pixel form a sample and are modified by the so-called local gradual descent matrix (LGDM), values of which are descending from center toward outside. LGDM visits each sample and causes small variations in pixel values of the sample in an attempt to shift the sample toward the correct Gaussian component center in the feature space. Thus, LGDM decides how much modification to the current sample is necessary for true categorization of the current pixel by later <formula formulatype="inline" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><tex Notation="TeX">$k$</tex></formula> -means. The motivation behind the proposed approach is twofold. First, a general method that could efficiently explore both local and global changes for unsupervised change detections is needed. Second, unsupervised change detection methods generally use nonsystematic selections of system parameters. Hence, a parameter selection method without using the ground truth image is required for unsupervised methods. The proposed change detection method is tested for both optical and advanced synthetic aperture radar satellite images and compared with the recent works based on the same input set. The proposed method outperforms the others qualitatively and quantitatively.

A Volumetric Approach to Change Detection in Satellite Images

The increasing availability of very high resolution satellite imagery has spurred interest in automatically detecting very fine detailed changes in an area over time, a particularly useful tool for analyzing activity in dense urban areas. However, attempting automated change detection at this resolution is difficult due to the motion parallax of elevated structures. This paper presents a comprehensive solution to change detection in areas of significant 3D relief using a new framework called volumetric appearance modeling (VAM). This approach can manage the complications of unknown and changing world surfaces by maintaining a 3D voxel-based model, where probability distributions for surface occupancy and image appearance are stored in each voxel. These distributions are continuously updated as new images are received using an adaptive learning procedure. This representation is demonstrated to produce accurate change detection results under conditions of variable illumination and viewpoint as well as haze conditions present in satellite imagery. The volumetric representation also supports automatic sensor model correction to align incoming imagery to a common geographic reference. This registration approach is demonstrated to achieve geo-positioning accuracy on the order of the ground sampling distance (GSD) or better.

Method for unsupervised change detection in satellite images

A Gaussian mixture model (GMM) and Bayesian inferencing (BI) based unsupervised change detection method in satellite images is presented. The data distribution of the difference image, which is computed from satellite images of the same scene acquired at different time instances, is modelled by using GMM. The components of the overall GMM are separated into two classes to model the data distributions of changed and unchanged pixels. The weights of the components in each class are used to estimate the a priori probability of each corresponding class. The final change detection is achieved by applying BI to classify each pixel of the difference image into one or two classes.

An integrated approach to flood hazard assessment on alluvial fans using numerical modeling, field mapping, and remote sensing

Millions of people in the western United States live near the dynamic, distributary channel networks of alluvial fans where flood behavior is complex and poorly constrained. Here we test a new comprehensive approach to alluvial-fan flood hazard assessment that uses four complementary methods: two-dimensional raster-based hydraulic modeling, satellite-image change detection, field-based mapping of recent flood inundation, and surficial geologic mapping. Each of these methods provides spatial detail lacking in the standard method and each provides critical information for a comprehensive assessment. Our numerical model simultaneously solves the continuity equation and Manning's equation (Chow, 1959) using an implicit numerical method. It provides a robust numerical tool for predicting flood flows using the large, high-resolution Digital Elevation Models (DEMs) necessary to resolve the numerous small channels on the typical alluvial fan. Inundation extents and flow depths of historic floods can be reconstructed with the numerical model and validated against field- and satellite-based flood maps. A probabilistic flood hazard map can also be constructed by modeling multiple flood events with a range of specified discharges. This map can be used in conjunction with a surficial geologic map to further refine floodplain delineation on fans. To test the accuracy of the numerical model, we compared model predictions of flood inundation and flow depths against field- and satellite-based flood maps for two recent extreme events on the southern Tortolita and Harquahala piedmonts in Arizona. Model predictions match the field- and satellite-based maps closely. Probabilistic flood hazard maps based on the 10 yr, 100 yr, and maximum floods were also constructed for the study areas using stream gage records and paleoflood deposits. The resulting maps predict spatially complex flood hazards that strongly reflect small-scale topography and are consistent with surficial geology. In contrast, FEMA Flood Insurance Rate Maps (FIRMs) based on the FAN model predict uniformly high flood risk across the study areas without regard for small-scale topography and surficial geology.

Sensitivity of the Landsat enhanced wetness difference index (EWDI) to temporal resolution

We investigated the sensitivity of the Landsat enhanced wetness difference index (EWDI) to temporal resolution in detecting forest harvesting and silvicultural activities in the Fundy Model Forest in southern New Brunswick. The severity of disturbance was underestimated in the multiyear EWDI because of regrowth and natural variability. In certain instances the severity of disturbance was overestimated with an annual EWDI; for example, a herbicide application in a young stand could have an annual wetness difference approximately equal to that of a forest clearcut. In general, we found annual EWDI differences to be more accurate than multiyear differences in detecting change. The sensitivity of satellite image change detection techniques to temporal resolution must be considered when national or international protocols are developed to provide input to forest monitoring or carbon accounting efforts.