Synthetic aperture radar (SAR) images often interfere with speckle artifacts that have a great impact on subsequent processing and analysis operations. To remove speckle artifacts, this article introduces a hybrid denoising approach by using a convolutional neural network (CNN) and consistent cycle spinning (CCS) in the nonsubsample shearlet transform (NSST) domain. First, we apply NSST to a noisy SAR image to gain low- and high-frequency coefficients. Second, we adopt a learned deep CNN model to eliminate the speckle noise in the low-frequency coefficients, which retains more contour information. Third, we employ CCS to enhance the high-frequency coefficients, which preserves more details of the original SAR image. Finally, we obtain the denoised image by using inverse NSST applied to the denoised coefficients. Compared with state-of-the-art algorithms, the results of the experiment indicate that our method not only achieves better speckle removal performance but also maintains more detailed information retention.

SAR Speckle Removal Using Hybrid Frequency Modulations

The accuracy in prediction of change of land cover is important in growing urbanization. The changes on the surface of the earth occur due to disasters, deforestation, change in course of the river, urbanization, etc. If the change detection of the earth’s surface is completed timely and accurately, then the connection and interaction between natural phenomena and humans are often better analyzed and understood as a result of which better management and use of resources can be done. Detection of change is most important for correct understanding of land cover changes with multi-temporal data. Main idea is to detect the regions that have changed in various images of an equivalent place taken at different timestamps. It has many growing applications in real world such as smart cities, disease treatment, and forestry. In this paper, a comparative study is made for the detection of change in remote sensing images using different machine learning approaches. Initially, the difference image is computed from the time stamped multispectral images. Then, feature reduction is employed with three approaches which use latent space, principal component, and a variation of it. The extracted components are then given to k-means clustering for detection of change, and change map is obtained. The results clearly show that the factor analysis combined with k-means detected the change effectively and also with less time when compared with other algorithms. 

Comparison of Feature Reduction Techniques for Change Detection in Remote Sensing

Synthetic-aperture radar (SAR) images are speckle corrupted at the initial stage of their acquisition. For this reason, despeckling of the images is required to preserve the finer details without losing them. But, quite often it was seen that even after despeckling, there persists residual noise at higher noise levels and obscured or weak edges. Therefore, to overcome these problems, in this paper, a method of enhancement using dynamic unsharp masking has been proposed for various noise levels. It uses the concept of variable weight factor which is calculated for a single pixel of the image. Dynamic unsharp masking controls the enhancement of the images adaptively such that there is no overshooting in images. The sharpening competency of the proposed method is then scrutinized and compared with conventional unsharp masking using an evaluation parameter called measure of enhancement (EME).

Enhancement of Synthetic-Aperture Radar (SAR) Images Based on Dynamic Unsharp Masking

The importance of remote sensing has ballooned in the recent decade, due to the increased need for its applications, from commercial and economic, to intelligence and military applications. Synthetic Aperture Radar (SAR), an important remote sensing instrument captures data that are inherently affected with speckle noise due to its coherent imaging, which makes working with SAR data complicated. Therefore, Synthetic Aperture Radar (SAR) images can benefit greatly from good speckle reduction or suppression techniques, making them easier to analyze and utilize their maximum potential. This paper gives a technical review of various speckle filtering approaches for Synthetic Aperture Radar Images (SAR), discussing the drawbacks of the older conventional methods, and the improvements the newer convolutional Neural Networks (CNN) based solutions bring to the table.

Speckle Reduction in SAR Images using CNN

The pore characteristics of coal reservoirs play a very important role in the migration of coalbed methane reservoirs, and its heterogeneity has been studied by a large number of scholars. However, there are few studies on lignite reservoirs. In this paper, three sets of lignite reservoirs in the Jiergalangtu depression are taken as the research object. The scanning electron microscope (SEM) images are converted from time-domain signals to frequency-domain signals through Fourier transformation, which quantitatively characterizes the heterogeneity of coal reservoirs. The variance of the value indicates that the three sets of coal seams NO.4Var<NO.5Var<NO.6Var. Through nitrogen adsorption experiment and fractal analysis, the pore continuity, surface roughness, pore structure, pore type and other aspects are studied. The lignite in the study area has high pore surface roughness and complex pore structure. Affected by various geological factors such as material composition, coal-forming environment, degree of evolution and other geological factors, lignite reservoirs in different layers and different mining areas show obvious heterogeneity.

https://journals.sagepub.com/doi/pdf/10.1177/01445987221099088

Pore characteristics and its heterogeneity of lignite reservoir in the Erlian Basin of Inner Mongolia, China

Synthetic Aperture Radar (SAR) images are mostly damaged by signal dependent speckle noise. It reduces potential image abstraction and perception. Therefore speckle suppression is considered as a major step to overcome these problems. However, consequences of despeckling are blurred image and weak edges which is therefore followed by the application of enhancement methods on SAR images in order to increase the contrast level of the images. This paper gives an insight technical review of different speckle filtering and enhancement approaches for speckled remote sensing images; conclusion and recommendations are drawn eventually.

Speckle Suppression and Enhancement Approaches for Processing of SAR Images: A Technical Review

Study of Micro-structures and their Relation with Occurrence of Mineral Matter in Ramagundam Coals, Godavari Basin, India: Implications on Coal and Hydrocarbon Industries

Synthetic Aperture Radar (SAR) images are speckle corrupted right at the initial stage of their acquisition due to repeated backscattering signals from the uniformly distributed target sites. Addition of speckle noise in the images causes poor visualization, image analysis and image translation. For this reason, despeckling of the image data is important to not only remove the speckle but also preserve the finer details without losing them. In this paper, a method of despeckling using local statistics-based adaptive filtering has been proposed which can be used in all kinds of speckle environment especially in higher speckle levels. The filter is adaptive in nature as the size of the filtering window which is being used in this method can be varied to despeckle those regions with fully developed speckle and protect minute details. The filtering competency of the proposed filter is then scrutinized and compared using various metrics like Peak Signal-to-Noise Ratio (PSNR) and Speckle Suppression Index (SSI) on different speckle variances.

Ankita Rai

Papers

Speckle Suppression and Enhancement Approaches for Processing of SAR Images: A Technical Review

Study of Micro-structures and their Relation with Occurrence of Mineral Matter in Ramagundam Coals, Godavari Basin, India: Implications on Coal and Hydrocarbon Industries

Enhancement of Synthetic-Aperture Radar (SAR) Images Based on Dynamic Unsharp Masking

Despeckling of Synthetic Aperture Radar (SAR) Images Using Local Statistics-Based Adaptive Filtering

Characterization of composite coal seam III–IV of Ramagundam coalfield, Godavari Basin (India): an attempt to evaluate CBM prospects