Change Detection Performance Research Articles

Changing flight heading during pass to enhance SAR change detection performance

Changing flight heading during pass to enhance SAR change detection performance

Remote Sensing Image Translation via Style-Based Recalibration Module and Improved Style Discriminator

Existing remote sensing change detection methods are heavily affected by seasonal variation. Since vegetation colors are different between winter and summer, such variations are inclined to be falsely detected as changes. In this letter, we proposed an image translation method to solve the problem. A style-based recalibration module is introduced to capture seasonal features effectively. Then, a new style discriminator is designed to improve the translation performance. The discriminator can not only produce a decision for the fake or real sample, but also return a style vector according to the channel-wise correlations. Extensive experiments are conducted on season-varying dataset. The experimental results show that the proposed method can effectively perform image translation, thereby consistently improving the season-varying image change detection performance. Our codes and data are available at https://github.com/summitgao/RSIT_SRM_ISD.

Heterogeneous Change Detection on Remote Sensing Data with Self-Supervised Deep Canonically Correlated Autoencoders

Change detection is a well-known topic of remote sensing. The goal is to track and monitor the evolution of changes affecting the Earth surface over time. The recently increased availability in remote sensing data for Earth observation and in computational power has raised the interest in this field of research. In particular, the keywords “multitemporal” and “heterogeneous” play prominent roles. The former refers to the availability and the comparison of two or more satellite images of the same place on the ground, in order to find changes and track the evolution of the observed surface, maybe with different time sensitivities. The latter refers to the capability of performing change detection with images coming from different sources, corresponding to different sensors, wavelengths, polarizations, acquisition geometries, etc. This thesis addresses the challenging topic of multitemporal change detection with heterogeneous remote sensing images. It proposes a novel approach, taking inspiration from recent developments in the literature. The proposed method is based on deep learning - involving autoencoders of convolutional neural networks - and represents an exapmple of unsupervised change detection. A major novelty of the work consists in including a prior information model, used to make the method unsupervised, within a well-established algorithm such as the canonical correlation analysis, and in combining these with a deep learning framework to give rise to an image translation method able to compare heterogeneous images regardless of their highly different domains. The theoretical analysis is supported by experimental results, comparing the proposed methodology to the state of the art of this discipline. Two different datasets were used for the experiments, and the results obtained on both of them show the effectiveness of the proposed method.

Application of weighted sample entropy in detection of abnormal signal

An improved sample entropy (SE) algorithm named weighted sample entropy(WSE) is proposed in this paper to optimize the problems existing in SE based on weighting. The principle of WSE algorithm is given, and the performance of it for dynamical change detection is analyzed using synthetic signal. The results show that WSE can accurately capture amplitude information and amplify the detection results of dynamical changes compared with SE and PE in the case of high complexity, but it requires a higher computation cost than the others. From all the analyses in this paper, we find that WSE has a better performance for dynamical change detection compared with the other three algorithms in abnormal amplitude.

Multiple Kernel Graph Cut for SAR Image Change Detection

Complementary information between two difference images (DI’s) has great contribution to improve change detection performances. Based on the effectiveness and flexibility of the multiple kernel learning (MKL) in information fusion, we develop a multiple kernel graph cut (MKGC) algorithm for synthetic aperture radar (SAR) image change detection. An energy function containing a weighted summation kernel is proposed for fusing the complementary information between the subtraction image and the ratio image. By iteratively minimizing the energy function, the kernel weights, region parameters and region labels are estimated automatically and optimally. Besides of it avoids modeling, MKGC also has a complete description of the changed areas and the strong noise immunity. Experiments on real GaoFen-3 SAR data set demonstrate the effectiveness of the MKGC algorithm, and illustrate that it is a good candidate for SAR image change detection.

Impaired visual working memory and reduced connectivity in undergraduates with a history of mild traumatic brain injury

Mild traumatic brain injury (mTBI), or concussion, accounts for 85% of all TBIs. Yet survivors anticipate full cognitive recovery within several months of injury, if not sooner, dependent upon the specific outcome/measure. Recovery is variable and deficits in executive function, e.g., working memory (WM) can persist years post-mTBI. We tested whether cognitive deficits persist in otherwise healthy undergraduates, as a conservative indicator for mTBI survivors at large. We collected WM performance (change detection, n-back tasks) using various stimuli (shapes, locations, letters; aurally presented numbers and letters), and wide-ranging cognitive assessments (e.g., RBANS). We replicated the observation of a general visual WM deficit, with preserved auditory WM. Surprisingly, visual WM deficits were equivalent in participants with a history of mTBI (mean 4.3 years post-injury) and in undergraduates with recent sports-related mTBI (mean 17 days post-injury). In seeking the underlying mechanism of these behavioral deficits, we collected resting state fMRI (rsfMRI) and EEG (rsEEG). RsfMRI revealed significantly reduced connectivity within WM-relevant networks (default mode, central executive, dorsal attention, salience), whereas rsEEG identified no differences (modularity, global efficiency, local efficiency). In summary, otherwise healthy current undergraduates with a history of mTBI present behavioral deficits with evidence of persistent disconnection long after full recovery is expected.

Automated building change detection with amodal completion of point clouds

When updating digital models of existing buildings, changes in the built environment are detected by comparing outdated BIMs with captured point clouds representing current conditions. Here we show that point cloud completion (i.e. automated filling-in of missing data) improves the accuracy of change detection. We perform point cloud completion using a hierarchical deep variational autoencoder (a type of artificial neural network) modified to include skip connections between the convolution and deconvolution layers. The resulting receiver operating characteristic curve shows that completion boosts change detection performance from a total area under the curve of 0.55 to 0.75. Completion achieves this by eliminating differences between the BIM and point cloud inputs that are a consequence of incompleteness while distilling the differences due to building change. We anticipate that automated change detection methods with resilience to imperfect data will become more critical as automated building analyses become increasingly abstracted from data collection.

Preparing for the Worst: Attention is Enhanced Prior to Any Upcoming Emotional or Neutral Stimulus

Do people allocate more or fewer attentional resources when preparing for negative emotional visual stimuli to appear? In three experiments (total N = 150), participants performed a change-detection task while expecting a neutral, threatening, disgusting, or joyful stimulus or no stimulus to appear at a fixed moment. Responses to an infrequent dot probe were faster when participants were expecting a distracting stimulus. Importantly, although only negative stimuli impaired change-detection performance, there was no difference between the preparation effect for threatening and neutral stimuli (Experiment 1) or disgusting and joyful stimuli (Experiment 3). The preparation effects were also unaffected by the participant’s anxiety level. Experiment 2 confirmed that the threatening images affected performance when the dot probe appeared after the image. These results suggest that the visual system increases alertness in response to any upcoming stimulus and further imply that the effects of emotional stimuli largely occur after, but not before, the stimuli appear.

Combining Radar and Optical Satellite Imagery with Machine Learning to Map Lava Flows at Mount Etna and Fogo Island

Lava flow mapping has direct relevance to volcanic hazards once an eruption has begun. Satellite remote sensing techniques are increasingly used to map newly erupted lava, thanks to their capability to survey large areas with frequent revisit time and accurate spatial resolution. Visible and infrared satellite data are routinely used to detect the distributions of volcanic deposits and monitor thermal features, even if clouds are a serious obstacle for optical sensors, since they cannot be penetrated by optical radiation. On the other hand, radar satellite data have been playing an important role in surface change detection and image classification, being able to operate in all weather conditions, although their use is hampered by the special imaging geometry, the complicated scattering process, and the presence of speckle noise. Thus, optical and radar data are complementary data sources that can be used to map lava flows effectively, in addition to alleviating cloud obstruction and improving change detection performance. Here, we propose a machine learning approach based on the Google Earth Engine (GEE) platform to analyze simultaneously the images acquired by the synthetic aperture radar (SAR) sensor, on board of Sentinel-1 mission, and by optical and multispectral sensors of Landsat-8 missions and Multi-Spectral Imager (MSI), on board of Sentinel-2 mission. Machine learning classifiers, including K-means algorithm (K-means) and support vector machine (SVM), are used to map lava flows automatically from a combination of optical and SAR images. We describe the operation of this approach by using a retrospective analysis of two recent lava flow-forming eruptions at Mount Etna (Italy) and Fogo Island (Cape Verde). We found that combining both radar and optical imagery improved the accuracy and reliability of lava flow mapping. The results highlight the need to fully exploit the extraordinary potential of complementary satellite sensors to provide time-critical hazard information during volcanic eruptions.

Local Similarity Siamese Network for Urban Land Change Detection on Remote Sensing Images

Change detection is an important task in the field of remote sensing. Various change detection methods based on convolutional neural networks (CNNs) have recently been proposed for remote sensing using satellite or aerial images. However, existing methods allow only the partial use of content information in images during change detection because they adopt simple feature similarity measurements or pixel-level loss functions to construct their network architectures. Therefore, when these methods are applied to complex urban areas, their performance in terms of change detection tends to be limited. In this article, a novel CNN-based change detection approach, referred to as a local similarity Siamese network (LSS-Net), with a cosine similarity measurement, was proposed for better urban land change detection in remote sensing images. To use content information on two sequential images, a new change attention map-based content loss function was developed in this study. In addition, to enhance the performance of the LSS-Net in terms of change detection, a suitable feature similarity measurement method, incorporated into a local similarity attention module, was determined through systemic experiments. To verify the change detection performance of the LSS-Net, it was compared with other state-of-the-art methods. The experimental results show that the proposed method outperforms the state-of-the-art methods in terms of the F1 score (0.9630, 0.9377, and 0.7751) and kappa (0.9581, 0.9351, and 0.7646) on the three test datasets, thus suggesting its potential for various remote sensing applications.

New Forms of Likelihood Ratio Test for SAR Change Detection

The Neyman-Pearson lemma, i.e., the likelihood ratio test and its generalized version, have been used for the development of the synthetic aperture radar (SAR) change detection methods. For detecting changes caused by targets on the ground such as vehicles, a target model, or at least certain assumptions concerning the targets, is always required for deriving a statistical hypothesis test. Without the prior knowledge on targets, it is difficult to make any assumption. An inappropriate assumption can degrade change detection performance significantly. To avoid this technical issue, some new forms of likelihood ratio test for SAR change detection are introduced in this paper. The proposed forms are shown to be flexible. They can be utilized to develop change detection methods for different types of data, e.g., data in scalar form, data in vector form, data represented in complex number, and data represented in real number. The flexibility of the proposed forms is also shown by the capability to implement change detection methods in the iterative and non-iterative ways. For the illustration purpose, a new change detection method is developed on one of the introduced forms and tested with TanDEM-X data measured in Karlshamn, Sweden in 2016.

Ordering for Communication-Efficient Quickest Change Detection in a Decomposable Graphical Model

A quickest change detection problem is considered in a sensor network with observations whose statistical dependency structure across the sensors before and after the change is described by a decomposable graphical model (DGM). Distributed computation methods for this problem are proposed that are capable of producing the optimum centralized test statistic. The DGM leads to the proper way to collect nodes into local groups equivalent to cliques in the graph, such that a clique statistic which summarizes all the clique sensor data can be computed within each clique. The clique statistics are transmitted to a decision maker to produce the optimum centralized test statistic. In order to further improve communication efficiency, an ordered transmission approach is proposed where transmissions of the clique statistics to the fusion center are ordered and then adaptively halted when sufficient information is accumulated. This procedure is always guaranteed to provide the optimal change detection performance, despite not transmitting all the statistics from all the cliques. A lower bound on the average number of transmissions saved by ordered transmissions is provided and for the case where the change seldom occurs the lower bound approaches approximately half the number of cliques provided a well behaved distance measure between the distributions of the sensor observations before and after the change is sufficiently large. We also extend the approach to the case when the graph structure is different under each hypothesis. Numerical results show significant savings using the ordered transmission approach and validate the theoretical findings.

Building change detection from multi-source remote sensing images based on multi-feature fusion and extreme learning machine

ABSTRACT In this paper, in order to improve the accuracy of multi-source remote sensing image building change detection, we propose a method based on multi-feature fusion (MFF) and extreme learning machine (ELM) training. Firstly, we fuse the spectral feature difference image (DI) and textural feature (grey level co-occurrence matrix) DI obtained by change vector analysis (CVA), morphological building index DI, and shape feature DI obtained by subtraction to construct the final DI. Secondly, the coarse change detection map obtained by selecting a threshold for the DI saliency map obtained by the use of the frequency-domain significance (FDS) method is pre-classified by the fuzzy c-means (FCM) clustering algorithm. Finally, the neighborhood features obtained from the original images and the feature images of the changed pixels (buildings) and the unchanged pixels in the coarse change map are extracted and used as reliable samples for the ELM training. By using the trained ELM classifier, undetermined pixels are further separated into changed and unchanged classes. Finally, we combine the ELM classification result and the preclassification result together to form the final building change map. Experiments on two real multi-source datasets show that the proposed method can result in a significant improvement in multi-source remote sensing image building change detection performance.

Change Detection Method of High Resolution Remote Sensing Image Based on D-S Evidence Theory Feature Fusion

Using high-resolution satellite image to detect change has been a hotspot in the field of remote sensing for a long time series. The change detection method combining feature extraction and machine learning could extract the change information effectively, but the manual sample selection is a huge workload for a wide range remote sensing images, and it is also difficult to ensure the accuracy of the pre-detection sample using a single difference image. Therefore, in this paper, a new method for change detection has been put forward based on multi-feature fusion of D-S evidence theory. In this approach, the texture difference image has calculated by structural similarity, because the difference image based on structural similarity plays a great role in change detection, which was verified in experiments. The difference images based on texture features and traditional spectral features are fused by D-S evidence theory, and texture features and spectral features have been fully utilized. Setting rules to select samples with high confidence based on pixels, and SLIC super-pixel segmentation has applied in order to improve further the credibility of the sample. Finally, the samples selected by SLIC segmentation optimization are sent to the classifier training to obtain the final result. The experimental results show that texture features play a very important role in the change detection of high-resolution remote sensing images, and D-S evidence theory could effectively fuse spectral texture features to improve the accuracy of change detection. The proposed method has high accuracy and good performance in change detection.

A Novel Metal Foreign Object Detection for Wireless High-Power Transfer Using a Two-Layer Balanced Coil Array with a Serial-Resonance Maxwell Bridge

In a wireless high-power transfer system with a distance of several tens of centimeters apart between the transmitter and receiver coils, one of the most challenging issues is to detect metallic foreign objects between the transmitter and receiver coils. The metallic foreign objects must be detected and removed since these reduce the transmission efficiency and cause heat generation of the transmitter and receiver. This paper presents two-layer symmetric balanced coil array so that if there are metallic foreign objects, it can be detected through the change of the inductance of the balanced coils. Since the balanced coil is composed of coils that are in a symmetrical relationship in position, there is no need for a reference coil, and interference between coils is reduced by dividing the coil into two layers. In addition, a novel serial-resonance Maxwell bridge circuit to improve the inductance change detection performance is presented in this paper. The proposed metallic foreign object detection system is implemented using two-layer balanced coil array with a serial-resonance Maxwell bridge and the experimental results show that voltage changes of hundreds of mV to several V occur when a metallic foreign object is inserted, so that even small metals such as clips can be detected.

Weakly Supervised Change Detection Based on Edge Mapping and SDAE Network in High-Resolution Remote Sensing Images

Change detection for high-resolution remote sensing images is more and more widespread in the application of monitoring the Earth’s surface. However, on the one hand, the ground truth could facilitate the distinction between changed and unchanged areas, but it is hard to acquire them. On the other hand, due to the complexity of remote sensing images, it is difficult to extract features of difference, let alone the construction of the classification model that performs change detection based on the features of difference in each pixel pair. Aiming at these challenges, this paper proposes a weakly supervised change detection method based on edge mapping and Stacked Denoising Auto-Encoders (SDAE) network called EM-SDAE. We analyze the difference in edge maps of bi-temporal remote sensing images to acquire part of the ground truth at a relatively low cost. Moreover, we design a neural network based on SDAE with a deep structure, which extracts the features of difference so as to efficiently classify changed and unchanged regions after being trained with the ground truth. In our experiments, three real sets of high-resolution remote sensing images are employed to validate the high efficiency of our proposed method. The results show that accuracy can even reach up to 91.18% with our method. In particular, compared with the state-of-the-art work (e.g., IR-MAD, PCA-k-means, CaffeNet, USFA, and DSFA), it improves the Kappa coefficient by 27.19% on average.

Optical Remote Sensing Image Change Detection Based on Attention Mechanism and Image Difference

This study presents a new end-to-end change detection network, called difference-enhancement dense-attention convolutional neural network (DDCNN), that is designed for detection of changes in the bitemporal optical remote sensing images. To model the internal correlation between high-level and low-level features, a dense attention method consisting of several up-sampling attention units is proposed. Both the up-sampling spatial and up-sampling channel attention are adopted by the unit. The unit, which can use high-level features with rich category information to guide the selection of low-level features, can use the spatial context information to capture the changed features of ground objects. Furthermore, DDCNN also pays attention to the differentiating features of the bitemporal images. By introducing a DE unit, each pixel is weighted and the features are selectively aggregated. The combination of dense attention and the DE unit improves the effectiveness of the network and its accuracy in extracting the change features. The effectiveness of the proposed approach is demonstrated via five challenge data sets. The experimental results show that DDCNN achieves new state-of-the-art change detection performance on these five challenging data sets. For the seasonal change detection data set in particular, compared with the best existing change detection model, the proposed method increases the F1 score and IoU by 2.96% and 5.17%, respectively; compared with the baseline method, our method improved 3.75% and 6.50% on the F1 score and IoU, respectively.

A near-real-time approach for monitoring forest disturbance using Landsat time series: stochastic continuous change detection

Forest disturbances greatly affect the ecological functioning of natural forests. Timely information regarding extent, timing and magnitude of forest disturbance events is crucial for effective disturbance management strategies. Yet, we still lack accurate, near-real-time and high-performance remote sensing tools for monitoring abrupt and subtle forest disturbances. This study presents a new approach called ‘Stochastic Continuous Change Detection (S-CCD)’ using a dense Landsat data time series. S-CCD improves upon the ‘COntinuous monitoring of Land Disturbance (COLD)’ approach by incorporating a mathematical tool called the ‘state space model’, which treats trends and seasonality as stochastic processes, allowing for modeling temporal dynamics of satellite observations in a recursive way. The quantitative accuracy assessment is evaluated based on 3782 Landsat-based disturbance reference plots (30 m) from a probability sampling distributed throughout the Conterminous United States. Validation results show that the overall accuracy (best F1 score) of S-CCD is 0.793 with 20% omission error and 21% commission error, slightly higher than that of COLD (0.789). Two disturbance sites respectively associated with wildfire and insect disturbances are used for qualitative map-based analysis. Both quantitative and qualitative analyses suggest that S-CCD achieves fewer omission errors than COLD for detecting those disturbances with subtle/gradual spectral change. In addition, S-CCD facilitates a better real-time monitoring, benefited by its complete recursive manner and a shorter lag for confirming disturbance than COLD (126 days vs. 166 days for alerting 50% disturbance events), and reached up to ~4.4 times speedup for computation. This research addresses the need for near-real-time monitoring and large-scale mapping of forest health and offers a new approach for operationally performing change detection tasks from dense Landsat-based time series.

The effect of healthy aging on change detection and sensitivity to predictable structure in crowded acoustic scenes

The auditory system plays a critical role in supporting our ability to detect abrupt changes in our surroundings. Here we study how this capacity is affected in the course of healthy ageing. Artifical acoustic ‘scenes’, populated by multiple concurrent streams of pure tones (‘sources’) were used to capture the challenges of listening in complex acoustic environments. Two scene conditions were included: REG scenes consisted of sources characterized by a regular temporal structure. Matched RAND scenes contained sources which were temporally random. Changes, manifested as the abrupt disappearance of one of the sources, were introduced to a subset of the trials and participants (‘young’ group N = 41, age 20-38 years; ‘older’ group N = 41, age 60-82 years) were instructed to monitor the scenes for these events. Previous work demonstrated that young listeners exhibit better change detection performance in REG scenes, reflecting sensitivity to temporal structure. Here we sought to determine: (1) Whether ‘baseline’ change detection ability (i.e. in RAND scenes) is affected by age. (2) Whether aging affects listeners’ sensitivity to temporal regularity. (3) How change detection capacity relates to listeners’ hearing and cognitive profile (a battery of tests that capture hearing and cognitive abilities hypothesized to be affected by aging). The results demonstrated that healthy aging is associated with reduced sensitivity to abrupt scene changes in RAND scenes but that performance does not correlate with age or standard audiological measures such as pure tone audiometry or speech in noise performance. Remarkably older listeners’ change detection performance improved substantially (up to the level exhibited by young listeners) in REG relative to RAND scenes. This suggests that the ability to extract and track the regularity associated with scene sources, even in crowded acoustic environments, is relatively preserved in older listeners.

A deeply supervised image fusion network for change detection in high resolution bi-temporal remote sensing images

Change detection in high resolution remote sensing images is crucial to the understanding of land surface changes. As traditional change detection methods are not suitable for the task considering the challenges brought by the fine image details and complex texture features conveyed in high resolution images, a number of deep learning-based change detection methods have been proposed to improve the change detection performance. Although the state-of-the-art deep feature based methods outperform all the other deep learning-based change detection methods, networks in the existing deep feature based methods are mostly modified from architectures that are originally proposed for single-image semantic segmentation. Transferring these networks for change detection task still poses some key issues. In this paper, we propose a deeply supervised image fusion network (IFN) for change detection in high resolution bi-temporal remote sensing images. Specifically, highly representative deep features of bi-temporal images are firstly extracted through a fully convolutional two-stream architecture. Then, the extracted deep features are fed into a deeply supervised difference discrimination network (DDN) for change detection. To improve boundary completeness and internal compactness of objects in the output change maps, multi-level deep features of raw images are fused with image difference features by means of attention modules for change map reconstruction. DDN is further enhanced by directly introducing change map losses to intermediate layers in the network, and the whole network is trained in an end-to-end manner. IFN is applied to a publicly available dataset, as well as a challenging dataset consisting of multi-source bi-temporal images from Google Earth covering different cities in China. Both visual interpretation and quantitative assessment confirm that IFN outperforms four benchmark methods derived from the literature, by returning changed areas with complete boundaries and high internal compactness compared to the state-of-the-art methods.