Analysis Of Synthetic Aperture Radar Images Research Articles

Statistically Principled Deep Learning for SAR Image Segmentation

This paper proposes a novel approach for Synthetic Aperture Radar (SAR) image segmentation by incorporating known statistical properties of SAR into deep learning models. We generate synthetic data using the Generalized Gamma distribution, modify the U-Net architecture to encompass statistical moments, and employ stochastic distance losses for improved segmentation performance. Evaluation against traditional methods will reveal the potential of this approach to advance SAR image analysis, with broader applications in environmental monitoring and general image segmentation tasks.

Automatic Speckle Noise Reduction in SAR Imagery Using Optimized Kernel Support Vector Machine

Speckle noise (SN) is one of the major types of noise that frequently occurs in different coherent imaging systems such as medical imaging, Synthetic Aperture Radar (SAR) and active Radar images. SAR is a powerful imaging technology that generates fine-resolution images and monitors the earth’s surface in order to identify its physical properties. The satellite images captured by SAR are mainly affected by SN, which reduces the quality of images and complicates the image representation. Therefore, removing SN from SAR images is one of the major challenges and needs significant attention. The proposed study introduces an optimal Machine Learning (ML) classifier named Kernel Support Vector Machine-Improved Aquila Optimization (KSVM-IAO) for reducing SN in SAR images. This study uses a two-step process called filtering and enhanced despeckling to minimize the consequence of speckle suppression. In the first step, different imaging filters, namely Improved Lee Filter (ILF), Improved Frost Filter (IFF), Improved Kuan Filter (IKF) and Improved Boxcar Filter (IBF), are utilized to remove the SN in SAR images. Next, the denoised image is fed to the second stage, which makes use of an optimized KSVM-IAO classifier to obtain an enhanced despeckle image. The hyperparameters of KSVM are tuned using the IAO algorithm, which reduces overfitting issues and increases accuracy. MATLAB is the simulation tool used for the analysis of SAR images. The simulation outcomes reveal that the proposed KSVM-IAO method obtained excellent SN removal while preserving the edges and fine details with low computational complexity.

Target Recognition Using Pre-Trained Convolutional Neural Networks and Transfer Learning

In modern surveillance, automatic target recognition (ATR) is a critical challenge, necessitating rapid and precise object identification, especially in military and disaster response scenarios. This research presents a comprehensive framework for ground target classification, focusing on Synthetic Aperture Radar (SAR) imagery. Harnessing the Moving and Stationary Target Acquisition and Recognition (MSTAR) Mixed Targets dataset, this study integrates Convolutional Neural Networks (CNNs) into SAR image analysis. Transfer learning emerges as a key strategy, adapting insights from pre-trained CNNs to the intricacies of target recognition in SAR imagery. This approach outperforms traditional methods, enhancing efficiency amid dataset annotation challenges. Noteworthy CNN architectures including DarkNet-19, DenseNet-121, InceptionV4, ResNet-152, and VGG19 are explored. The ResNet-152 model demonstrates exceptional performance, emerging as a leading contender with a remarkable testing accuracy of 98.56% when trained from scratch. Furthermore, by employing transfer learning, the model’s accuracy may be further enhanced to reach 98.81% on the testing dataset. This research signifies a transformative SAR imagery path guided by pre-trained CNNs and transfer learning. It reshapes ATR’s accuracy and efficiency, pointing to a future where innovation surmounts challenges previously deemed insurmountable.

Weighted residual network for SAR automatic target recognition with data augmentation.

Decades of research have been dedicated to overcoming the obstacles inherent in synthetic aperture radar (SAR) automatic target recognition (ATR). The rise of deep learning technologies has brought a wave of new possibilities, demonstrating significant progress in the field. However, challenges like the susceptibility of SAR images to noise, the requirement for large-scale training datasets, and the often protracted duration of model training still persist. This paper introduces a novel data augmentation strategy to address these issues. Our method involves the intentional addition and subsequent removal of speckle noise to artificially enlarge the scope of training data through noise perturbation. Furthermore, we propose a modified network architecture named weighted ResNet, which incorporates residual strain controls for enhanced performance. This network is designed to be computationally efficient and to minimize the amount of training data required. Through rigorous experimental analysis, our research confirms that the proposed data augmentation method, when used in conjunction with the weighted ResNet model, significantly reduces the time needed for training. It also improves the SAR ATR capabilities. Compared to existing models and methods tested, the combination of our data augmentation scheme and the weighted ResNet framework achieves higher computational efficiency and better recognition accuracy in SAR ATR applications. This suggests that our approach could be a valuable advancement in the field of SAR image analysis.

A conventional and non-conventional analysis of SAR image despeckling technique

The importance of high-resolution Synthetic Aperture Radar (SAR) imaging is undeniable in applications for Earth monitoring since it offers useful data for analysis. Post-image processing methods including edge and object detection, segmentation, and speckle noise removal are important to performed to find the information of images. The key technique that makes images aesthetically pleasing and understandable is despeckling. The presented study assays the latest trends & state of the art techniques for SAR image despeckling to gauge the performance & agility of existing techniques by performing a contemporary combination of theoretical & experimental analysis. This study analyses various techniques used in literature. The quantitative & qualitative analysis is done using Peak Signal-to-Noise Ratio (PSNR) and Universal Image Quality Index (UIQI) indices to find the better approach

Synthetic Aperture Radar image analysis based on deep learning: A review of a decade of research

Artificial intelligence research in the area of computer vision teaches machines to comprehend and interpret visual data. Machines can properly recognize and classify items using digital images captured by cameras and videos, deep learning models, and then respond to what they observe. Similarly, artificial intelligence has also been able to learn complex images captured by Synthetic Aperture Radar (SAR) that are widely used for various purposes but still leave room for improvements. Researchers have proposed numerous approaches in this field, from SAR target detection to SAR target recognition. This paper presents a survey on the different techniques and architectures proposed in the literature for various SAR image applications. The paper covers a survey on target detection models and target recognition models and their respective workflow to analyze the techniques involved and the performances of these models. This paper makes novel discussions, comparisons, and observations. It highlights the advantages and disadvantages of different approaches to give researchers the idea of how each technique can influence the performance for adoption in the future. The potential future directions along with hybrid models on each processing method are also highlighted based on the study.

Fundamental Study on Grasping Growth State of Paddy Rice Using Quad-Polarimetric SAR Data

In this brief paper, we examine polarimetric scattering characteristics for understanding seasonal change of paddy rice growth by using quad-polarimetric synthetic aperture radar (SAR) data in the X-band. Here we carry out polarimetric scattering measurement for a simplified paddy rice model in an anechoic chamber at X-band frequency to acquire the the quad polarimetric SAR data from the model. The measurements are performed several times for each growth stage of the paddy rice corresponding to seasonal change. The model-based scattering power decomposition is used for the examination of polarimetric features of the paddy rice model. It is found from the result of the polarimetric SAR image analysis for the measurement data that the growth state of the paddy rice in each stage can be understood by considering the ratio of the decomposition powers, when the planting direction of the paddy rice is not only normal but also oblique to radar direction. We can also see that orientation angle compensation (OAC) is useful for improving the accuracy of the growth stage observation in late vegetative stage for oblique planting case.

Progress in small object detection for remote sensing images

独特的拍摄视角和多变的成像高度使得遥感影像中包含大量尺寸极其有限的目标，如何准确有效地检测这些小目标对于构建智能的遥感图像解译系统至关重要。本文聚焦于遥感场景，对基于深度学习的小目标检测进行全面调研。首先，根据小目标的内在特质梳理了遥感影像小目标检测的3个主要挑战，包括特征表示瓶颈、前背景混淆以及回归分支敏感。其次，通过深入调研相关文献，全面回顾了基于深度学习的遥感影像小目标检测算法。选取3种代表性的遥感影像小目标检测任务，即光学遥感图像小目标检测、SAR图像小目标检测和红外图像小目标检测，系统性总结了3个领域内的代表性方法，并根据每种算法使用的技术思路进行分类阐述。再次，总结了遥感影像小目标检测常用的公开数据集，包括光学遥感图像、SAR图像及红外图像3种数据类型，借助于3种领域的代表性数据集SODA-A（small object detection datasets）、AIR-SARShip和NUAA-SIRST（Nanjing University of Aeronautics and Astronautics，single-frame infrared small target），进一步对主流的遥感影像目标检测算法在面对小目标时的性能表现进行横向对比及深入评估。最后，对遥感影像小目标检测的应用现状进行总结，并展望了遥感场景下小目标检测的发展趋势。;Remote sensing images are often captured from multiview and multiple altitudes, thereby comprising a mass of objects with limited sizes which significantly challenge current detection methods that can achieve outstanding performance on natural images. Moreover, how to precisely detect these small objects plays a crucial role in developing an intelligent interpretation system for remote sensing images. Focusing on the remote sensing images, this paper conducts a comprehensive survey for deep learning-based small object detection(SOD) can be reviewed and analyzed literately, including 1) features-represent bottlenecks, 2) background confusion, and 3) branching-regressed sensitivity. Specially, one of the major bottlenecks is for objective's representation. It refers that the down-sample operations in the prevailing feature extractors can suppress the signals of small objects unavoidably, and the following detection is impaired further in terms of the weak representations. The detection of size-limited instances is also interference of the confusion between the objects and backgrounds and the sensitivity of regression branch. For the former, the representations of small objects tend to be contaminated in related to feature extraction-contextual factors, which may erase the discriminative information that plays a significant role in head network. And the sensitivity of regression branch in small object detection is derived from the low tolerance for bounding box perturbation, in which a slight deviation of a predicted box will cause a drastic drop on the intersection-over-union(IoU), which is generally adopted to evaluate the accuracy of localization. Furthermore, we review and analyze the literature of small object detection for remote sensing images in the deep-learning era. In detail, by systematically reviewing corresponding methods of three small object detection tasks, i. e., SOD for optical remote sensing images, SOD for synthetic aperture radar(SAR) images and SOD for infrared images, an understandable taxonomy of the reviewed algorithms for each task is given. Specifically, we rigorously split the representative methods into several categories according to the major techniques used. In addition to the algorithm survey, considering the deep learning-based methods are hungry for data and to provide a comprehensive survey about small object detection, we also retrospect several publicly available datasets which are commonly used in these three SOD tasks. For each concrete field, we list the prevailing benchmarks in accordance with the published papers, and a brief introduction and some example images about these datasets are illustrated:small size. What is more, other related features about small object detection are highlighted as well, such as image resolution, data source, the number of images and annotated instances, and some proper statistics of each task, etc. Additionally, to better investigate the performance of generic detection methods on small objects, we analyze an in-depth evaluation and comparison of main-stream detection algorithms and several SOD methods for remote sensing images, namely SODA-A (small object detection datasets), AIR-SARShip and NUAA-SIRST (Nanjing University of Aeronautics and Astronautics, single-frame infrared small target). Afterwards, current situation in applications of small object detection for remote sensing images are analyzed, including SOD-based intelligent transportation system and scene-related understanding. Such harbor-targeted recognition is based on SAR image analysis, the precision-guided weapons based on the detection and recognition techniques of infrared images, and the tracking of moving targets at sea on top of multimodal remote sensing data. In the end, to enlighten the further research of small object detection in remote sensing images, we discuss four promising directions in the future. Concretely, it is required that an efficient backbone network can avoid the information loss of small objects while capturing the discriminative features to optimize the down-stream tasks about small objects. Large-scale benchmarks with well annotated small instances play an irreplaceable role linked to small object detection in remote sensing images further. Moreover, a multimodal remote sensing data-collaborated SOD algorithm is also preferred. A proper evaluation metric can not only guide the training and inference of small object detection methods under some specific scenes, but also rich its domain-related development.

Analyzing of land subsidence by Sentinel-1 time-series images using PSInSAR method: A case study of Thai Nguyen, Vietnam

Natural disasters and human activities are now causing an increase in land subsidence, or surface displacement. The effects of land subsidence cause landslides and construction cracking. PSInSAR (Persistent Scatter Interferometry) was identified to estimate surface displacement from a time series of Synthetic Aperture Radar (SAR) images. This technique is a subset of the DInSAR (Differential Interferometric Synthetic Aperture Radar) method. In this article, the authors determined and analyzed land subsidence in Thai Nguyen Province, Vietnam, using time series Sentinel-1A data with VV polarization from July 2019 to December 2020 and the PSInSAR method. There are numerous mineral exploitation mines in Thai Nguyen Province. It is one of the causes of an unusual amount of land subsidence in the region. According to the results determined by the InSAR technique, the velocity of displacement along the line of sight (LOS) of the study area ranges from -23.2 mm per year to +21.0 mm per year. The analysis of time-series SAR images reveals anomalous land subsidence at persistent scatter (PS) points. By analyzing the time-series displacement at PS points using the StaMPS Visualizer tool, the land subsidence during the image acquisition period and surface displacement trends over time were determined. According to this, coal mining regions have the highest land subsidence values ranging from -40 mm to -60 mm. The city and mine regions of Thai Nguyen, where operations have stopped, are largely stable. In addition, the time-series analysis at PS points will allow us to identify unusual displacement points, enabling the implementation of early warning plans.

Optimal parameters selection of the Frost filter based on despeckling efficiency prediction for Sentinel SAR images

Synthetic aperture radar (SAR) images have found numerous applications. However, further analysis of SAR images including interpretation, classification, segmentation, etc. is an extremely challenging task due to the presence of highly intensive speckle noise. Therefore, image despeckling is one of the main stages in preliminary SAR data processing. Over the past decades, a large number of different image despeckling techniques have been proposed ranging from local statistics filters to deep learning based ones. In this study, we analyze one of the most known and widely used local statistics Frost filter. Despeckling efficiency of the Frost filter significantly depends on the sliding window size and tuning (also called damping) factor. Here, we present a method for optimal parameters selection of the Frost filter for a given image based on despeckling efficiency prediction. Despeckling efficiency prediction for the Frost filter is carried out using a set of statistical and spectral input parameters and multilayer neural network. It is shown that such a prediction can be performed before applying image despeckling with a high accuracy and it is faster than despeckling itself. Both simulated speckled images and real-life Sentinel-1 SAR images have been used for extensive evaluation of the proposed method.

SAR Object Detection Using a Novel Convolutional Neural Network Architecture

In recent years, the importance of SAR (synthetic aperture radar) image analysis is growing day by day due to the vast applications in the field of oceanography, military war, land observation, agriculture, disaster management and geography. These applications required accurate classification of highresolution SAR images. In the present study, two different machine learning techniques are applied on the SAR dataset to analyze the classification accuracy. The first classification technique is SVM (support vector machine) along with principal component analysis (PCA) for feature reduction. While, in the second technique, a novel CNN (convolutional neural network) has been proposed to perform the classification on SAR images. Essential experimental analysis has been carried out based on the MSTAR dataset. Experimental results of the current study showed that the proposed CNN model outperformed the SVM classifier with an accuracy of 98.69%

Multi-Class Dual- Stream Convolutional Neural Network for Synthetic Aperture Radar Automatic Target Recognition of Ground Military Vehicle

Synthetic Aperture Radar (SAR) target classification is one of the largest branches of SAR image analysis. Despite the remarkable achievements of deep learning-based SAR target prediction algorithms, current object recognition algorithms are limited in terms of military applications. Acquisition and labeling of SAR target images are time-consuming and cumbersome. Obtaining adequate training data is also challenging in many cases. Deep learning-based models are always susceptible to overfitting because of insufficient training data. This limitation prevents them from being widely used to classify SAR targets. To overcome the problem of insufficient sampling and to learn more accurate representations for SAR image recognition, we propose a two way input of SAR images into a dual stream of DCNN. Concatenating two input SAR images representations was done using the restricted raw SAR data in order to extract the integral features from the 2 input SAR images representations for classification. The proposed methodology addressed the problem of insufficient sample in SAR target classification and improved classification accuracy without overfitting. Experimental results confirmed that the proposed method is effective in addressing the problem of insufficient sample in SAR target classification. This technique can be integrated into any SAR classification model based on convolutional neural networks (CNNs). The model MCDS-CNN results in a 99.1% recognition accuracy. Despite the limited availability of SAR image data from MSTAR, this approach provides good recognition results.

Extraction and Mitigation of Radio Frequency Interference Artifacts Based on Time-Series Sentinel-1 SAR Data

Radio frequency interference (RFI) is a critical issue for accurate remote sensing by synthetic aperture radar (SAR). Existing literature mainly detects and mitigates RFI in the raw data domain, which is generally not accessible to the end-user. In this article, a novel RFI extraction and mitigation scheme in the image domain is proposed using multitemporal analysis of SAR images. By exploiting the coupling correlation and complementary information among the time-series images, the background landscape could be modeled as relatively stationary with the low-rank property. Meanwhile, the radiometric artifacts corresponding to RFI could be well extracted and characterized by the sparse components. Extraction and mitigation of RFI signatures could be achieved simultaneously via a joint iterative optimization process. Experimental results on typical real-measured Sentinel-1 datasets acquired in different regional areas with various RFI types demonstrate the validity of the proposed method.

Diffusion of gravity waves by random space inhomogeneities in pancake-ice fields. Theory and validation with wave buoys and synthetic aperture radar

We study the diffusion of ocean waves by ice bodies much smaller than a wavelength, such as pancakes and small ice floes. We argue that inhomogeneities in the ice cover at scales comparable to that of the wavelength significantly increase diffusion, producing a contribution to wave attenuation comparable to what is observed in the field and usually explained by viscous effects. The resulting attenuation spectrum is characterized by a peak at the scale of the inhomogeneities in the ice cover, which could explain the rollover of the attenuation profile at small wavelengths observed in field experiments. The proposed attenuation mechanism leads to the same behaviors that would be produced by a viscous wave model with effective viscosity linearly dependent on the ice thickness. This may explain recent findings that viscous wave models require a thickness-dependent viscosity to fit experimental attenuation data. Experimental validation is carried out using wave buoy attenuation data and synthetic aperture radar image analysis.

Discrimination and Correlation Analysis of Multiview SAR Images with Application to Target Recognition

A multiview synthetic aperture radar (SAR) target recognition with discrimination and correlation analysis is proposed in this study. The multiple views are first prescreened by a support vector machine (SVM) to select out those highly discriminative ones. These views are then clustered into several view sets, in which images share high correlations. The joint sparse representation (JSR) is adopted to classify SAR images in each view set, and all the decisions from different view sets are fused using a linear weighting strategy. The proposed method makes more sufficient analysis of the multiview SAR images so the recognition performance can be effectively enhanced. To test the proposed method, experiments are set up based on the moving and stationary target acquisition and recognition (MSTAR) dataset. The results show that the proposed method could achieve superior performance under different situations over some compared methods.

Ridgelet-Nets With Speckle Reduction Regularization for SAR Image Scene Classification

With powerful feature representations, convolutional neural networks (CNNs) have produced tremendous achievements in image classification tasks and, typically, entail millions of labeled samples to train massive parameters. However, the sample labeling of synthetic aperture radar (SAR) images is extremely difficult, especially pixelwise labels, and has, sometimes, required field trips to accomplish labeling. Moreover, the inherent speckle noise may weaken the ability of networks to extract effective features from SAR images. In this article, we address these issues by labeling a few patchwise samples and propose Ridgelet-Nets with speckle reduction regularization for SAR image scene classification by combining deep learning with multiscale geometric analysis and statistical modeling of SAR images. First, we design Ridgelet-Nets with convolutional kernels constructed by ridgelet filters to reduce the training parameters and learn more discriminative features. Then, we embed speckle reduction regularization in the Ridgelet-Nets to restrain the influence of speckle noise and smooth the classification maps, in which the prior information of SAR image statistical modeling is introduced. Finally, we propose an adaptive SAR image scene classification framework based on an extended hierarchical visual semantic model, considering the differences in the structures and spatial relationships of different regions in the SAR images, particularly large-scale and complex scenes. Experimental results on real SAR images demonstrate that the proposed framework can achieve preferable classification performance using very limited labeled samples.

Semantic Labeling of Globally Distributed Urban and Nonurban Satellite Images Using High-Resolution SAR Data

While the analysis and understanding of multispectral (i.e., optical) remote sensing images have made considerable progress during the last decades, the automated analysis of synthetic aperture radar (SAR) satellite images still needs some innovative techniques to support nonexpert users in the handling and interpretation of these big and complex data. In this article, we present a survey of existing multispectral and SAR land cover image datasets. To this end, we demonstrate how an advanced SAR image analysis system can be designed, implemented, and verified that is capable of generating semantically annotated classification results (e.g., maps) as well as local and regional statistical analytics such as graphical charts. The initial classification is made based on Gabor features and followed by class assignments (labeling). This is followed by inclusion. This can be accomplished by the inclusion of expert knowledge via active learning with selected examples, and the extraction of additional knowledge from public databases to refine the classification results. Then, based on the generated semantics, we can create new topic models, find typical country-specific phenomena and distributions, visualize them interactively, and present significant examples including confusion matrices. This semiautomated and flexible methodology allows several annotation strategies, the inclusion of dedicated analytics procedures, and can generate broad as well as detailed semantic (multi-)labels for all continents, and statistics or models for selected countries and cities. Here, we employ knowledge graphs and exploit ontologies. These components could already be validated successfully. The proposed methodology can also be adapted to other SAR instruments with different resolutions as well as to multispectral images.

SPB-Net: A Deep Network for SAR Imaging and Despeckling With Downsampled Data

Synthetic aperture radar (SAR) typically faces both large-scale data and speckle noise problems, which, respectively, induce enormous strains on transmission and storage and interfere with the analysis and interpretation of SAR images. To tackle these, we present a real-time processing deep network, called SPB-Net, to implement imaging and speckle suppression simultaneously. First, a novel imaging-despeckling observation model with the nonlogarithmic additive speckle noise is established. Subsequently, guided by the statistical properties of noise and sparse and detail-preserved requirements in SAR imaging and despeckling, we formulate an <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$L_{2}$ </tex-math></inline-formula> along with two <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$L_{1}$ </tex-math></inline-formula> regularizations as the fidelity, sparse, and image detail-preserved constraints, respectively. Convolution layers are employed to improve the feature representation capability in the latter <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$L_{1}$ </tex-math></inline-formula> term as well. Based on this, we construct a corresponding convex optimization problem. Then, the complex-valued split Bregman method, focusing on the complex-variable convex problem, is unfolded into a parameter-learnable and architecture-fixed SPB-Net to solve the proposed problem effectively and efficiently. Experimental results with the downsampled Radarsat-1 raw data demonstrate the validity in imaging and speckle suppression and the real-time processing capability of the proposed SPB-Net.

Assessment of Surface Hydrological Connectivity in an Ungauged Multi-Lake System with a Combined Approach Using Geostatistics and Spaceborne SAR Observations

Connectivity metrics for surface water are important for predicting floods and droughts, and improving water management for human use and ecological integrity at the landscape scale. The integrated use of synthetic aperture radar (SAR) observations and geostatistics approach can be useful for developing and quantifying these metrics and their changes, including geostatistical connectivity function (GCF), maximum distance of connection (MDC), surface water extent (SWE), and connection frequency. In this study, we conducted a geostatistical analysis based on 52 wet and dry binary state (i.e., water and non-water) rasters derived from Sentinel-1 A/B GRD products acquired from 2015 to 2019 for China’s Momoge National Nature Reserve to investigate applicability and dynamics of the hydrologic connectivity metrics in an ungauged (i.e., data such as flow and water level are scarce) multi-lake system. We found: (1) generally, the change of GCF in North–South and Northeast–Southwest directions was greater than that in the West–East and Northwest–Southeast directions; (2) MDC had a threshold effect, generally at most 25 km along the W–E, NW–SE and NE–SW directions, and at most 45 km along the N–S direction; (3) the flow paths between lakes are diverse, including channelized flow, diffusive overbank flow, over-road flow and “fill-and-merge”; (4) generally, the values of the three surface hydrological connectivity indicators (i.e., the MDC, the SWE, and the conneciton frequency) all increased from May to August, and decreased from August to October; (5) generally, the closer the distance between the lakes, the greater the connection frequency, but it is also affected by the dam and road barrier. The study demonstrates the usefulness of the geostatistical method combining Sentinel-1 SAR image analysis in quantifying surface hydrological connectivity in an ungagged area. This approach should be applicable for other geographical regions, in order help resource managers and policymakers identify changes in surface hydrological connectivity, as well as address potential impacts of these changes on water resources for human use and/or ecological integrity at the landscape level.

Deep Learning Semantic Statistical Features’ Labeling Of SAR Image Based On K-MEANS Clustering And Wavelet Transform

[Purpose] Because Synthetic Aperture Radar (SAR) image contains much speckle, at present, it’s hard to obtain good results for SAR image with the traditional image classification methods. However, SAR image has rich texture information which facilitates SAR image segmentation. There is a Deep Learning method of K-means data clustering that is used to calculate the optimal k value to avoid more-segmentation or less-segmentation, the advantages of fuzzy theory and neural network are used to improve the classification accuracy of image processing.[Methods] According to the statistical characteristics of SAR images and the semantics of fuzzy neural networks analysis, an efficient image segmentation method based on Deep Learning Semantic analysis and wavelet transform is proposed to achieve precision of classification; Firstly, the texture features of SAR image are extracted by Deep Learning Semantic clustering; Secondly, the SAR image are characterized according to SAR semantics. Based on one of the Deep Learning Semantic k-means algorithms, the best k-value iterations of SSE(Sum of the Squared Errors)and SC(Silhouette Coefficient) are performed, and the suitable value k is selected; Finally, by Deep Learning segmentation based on the texture feature of the SAR image and the filtered gray component vector, and the SAR image is classified to facilitate the change detection of images.[Results] The experimental results show that a good k value detection is performed on the CPU/GPU platform. The SAR images is compared with two different results between the wavelet filtering and Deep Learning semantic method images which are single or multiple standard classified through detection-change SAR is better than the Empirical Approach, then the labels are classified for the accuracy and computational efficiency that are calculated. The classification results were accurately ameliorated. [Conclusions] The similarity calculation of k-value cluster is an important precondition, so it’s necessary to select the optimal k-value suitable for the segmentation condition. The method achieves are taken an adequate SAR image analysis of effect and improvement for performance by semantic classification of k-means clustering.