α-cut induced Fuzzy Deep Neural Network for change detection of SAR images

In this paper, a feature extraction approach based on a three-dimensional shearlet transform (shearlet 3D) is proposed. We aim at exploiting shearlet 3D to highlight the intrinsic properties of hyperspectral images (HSIs), well known by their correlated information and high dimensionality. First, we decompose the HSI to yield coefficients arranged in cubes that help the computing of statistical parameters. Afterward, using a simultaneous orthogonal matching pursuit (SOMP) algorithm, a classification process is carried out. SOMP relies on the powerful sparse representation paradigm, which helps representing data in a low-dimensional space. It is also built over the assumption that the contextual information incorporation into the sparse recovery problem improves the classification performance. To this algorithm, we propose to add an adapted decision rule where a similarity measurement is calculated to well assign the appropriate labels to pixels of interest. Experimental results proved that our proposed method outperformed state-of-the-art classifiers. Thanks to our proposed approach, we succeeded to build discriminative descriptors reaching high overall accuracies for two different HSI datasets, without taking into account all the shearlet 3D coefficients.

3-D Shearlet Transform Based Feature Extraction for Improved Joint Sparse Representation HSI Classification

Since its impoundment in 2003, the Three Gorges Reservoir (TGR) has brought far-reaching economic and other benefits to China. However, the resulting negative ecological and environmental impacts such as eutrophication and algal bloom have been problematic. This paper describes a study on nutrient estimation concerning eutrophication and algal bloom based on Huan Jing-1 (HJ-1) satellite imagery of the Xiangxi Bay (XXB), a tributary of the TGR in China. Models of nutrient concentration were established by correlation analysis between field measurement of nutrient concentration and the values obtained from the simultaneous satellite images of the study area. The results showed that total nitrogen (TN) had the best correlation with B1/(B3 + B4), with a maximal Spearman correlation coefficient of 0.885. Both nitrate (NO3
 −–N) and dissolved silicon (D-Si) had the best correlation with B1/(B2 + B4), with the maximal Spearman correlation coefficients of 0.818 and 0.756, respectively. All band combinations had little correlation with NH4
 +–N, TP, and PO4
 3−–P. The single variable quadratic models were established for predicting the concentration of TN and D-Si, while, a single variable cubic model was for the concentration of NH4
 +–N. Robustness analysis and validation were undertaken for the feasibility of the models, which showed that the predicted models were acceptable. Through the models, the levels of TN and NO3
 −–N were predicted on Jan 12, May 2, July 30 and Oct 8, 2010. The results showed that the water in the XXB was eutrophic mostly. The nitrogen level was higher on Jan 12 than that of May 2, July 30 and Oct 8. Compared with the upstream, the downstream of the XXB had higher nitrogen level on Jan 12 and May 2, whereas, on July 30, it was higher for the upstream. The nitrogen level was lowest on Oct 8, with the uniform distribution in the upstream and downstream of the XXB. The predicted results coincided with the field observation. The results will help to estimate the eutrophication and algal bloom through large-scale remote sensing monitoring and water quality management in the TGR.

Nutrient estimation by HJ-1 satellite imagery of Xiangxi Bay, Three Gorges Reservoir, China

Synthetic Aperture Radar (SAR), as a microwave sensor that can sense a target all day or night under all-weather conditions, is of great significance for detecting water resources, such as coastlines, lakes and rivers. This paper reviews literature published in the past 30 years in the field of water body extraction in SAR images, and makes some proposals that the community working with SAR image waterbody extraction should consider. Firstly, this review focuses on the main ideas and characteristics of traditional water body extraction on SAR images, mainly focusing on traditional Machine Learning (ML) methods. Secondly, how Deep Learning (DL) methods are applied and optimized in the task of water-body segmentation for SAR images is summarized from the two levels of pixel and image. We also pay more attention to the most popular networks, such as U-Net and its modified models, and novel networks, such as the Cascaded Fully-Convolutional Network (CFCN) and River-Net. In the end, an in-depth discussion is presented, along with conclusions and future trends, on the limitations and challenges of DL for water-body segmentation.

/pdf/water-body-segmentation-for-sar-images-past-current-and-11g7ezk4.pdf

Water-Body Segmentation for SAR Images: Past, Current, and Future

Taxonomic structure of phytoplankton in Shershnevskoe Reservoir (Chelyabinsk, Russia), an artificial lake

A novel change detection method based on high-resolution SAR images for river course

The purpose of this study was to elucidate the role of nutrient loading played in the eutrophication of newly built reservoirs Taking the Laohutan Reservoir as an example, field monitoring and laboratory experiments were conducted before and after the reservoir impoundment According to the results, the Laohutan Reservoir was eutrophicated 4 months after its impoundment, since nutrient loading played an important role Riverine inputs were found to be the major source of nutrients, with the contribution of greater than 72 % Nutrients from soil were also influential during the first 20 days Its release was gradually weakened by the routine operations of the reservoir The release patterns of nutrients from the four main land-use types of soil (forest, bamboo groves, paddy and construction sites) were similar, but their influence to the reservoir water quality varied dramatically And the atmosphere was also an important source of nutrients, especially nitrogen Meanwhile, by a comparison, the Laohutan Reservoir to an older one located nearby, the newer with relatively lower trophic levels underwent a cyanobacteria bloom, while there was none in the older It verified that new reservoirs were more prone to eutrophication than the older, and high nutrient loading is a necessary but not sufficient condition for eutrophication The fragility of the new reservoir’s ecosystem accelerated water environment deterioration, and resulted in the eutrophication and cyanobacteria bloom during the initial impoundment period

Eutrophication in the newly built Laohutan Reservoir during the initial impoundment period: the role of nutrient loading

The invention brings forward a multiple-dimensioned offshore oil-spill SAR image segmentation method based on JSEG and spectrum clustering, for solving the disadvantage of poor anti-noise performance of a conventional gray scale characteristic based spectrum cluster algorithm, and the advantages of single scale, sensitive directivity, optimum characteristic combination selection difficulty and the like of a conventional gray scale symbiosis matrix based multi-characteristic-spectrum cluster segmentation method. The method comprises: first of all, extracting the scale, spectrum and texture characteristics of an original image by use of a J-image multiple-dimensioned image sequence brought forward by a JSEG algorithm, constructing a spectrum cluster characteristic matrix, according to a normalized-cut criterion, using a K mean value cluster method to perform clustering on a Laplace matrix, and finally performing decision-level fusion on a segmentation result with a single scale by use of a ballot mechanism based fusion strategy so as to realize SAR image oil-spill multiple-dimensioned segmentation. Compared to the conventional multi-characteristic-spectrum cluster method, it is proved by multiple groups of experiment results that the method provided by the invention has higher segmentation precision and quite high robustness.

Dongmei Li

Papers

A novel change detection method based on high-resolution SAR images for river course

Eutrophication in the newly built Laohutan Reservoir during the initial impoundment period: the role of nutrient loading

Multiple-dimensioned offshore oil-spill SAR image segmentation method based on JSEG and spectrum clustering