scholarly journals Refined UNet V4: End-to-End Patch-Wise Network for Cloud and Shadow Segmentation with Bilateral Grid

2022 ◽  
Vol 14 (2) ◽  
pp. 358
Author(s):  
Libin Jiao ◽  
Lianzhi Huo ◽  
Changmiao Hu ◽  
Ping Tang ◽  
Zheng Zhang
Keyword(s):  
Remote Sensing ◽  
Message Passing ◽  
Shadow Detection ◽  
Landsat 8 ◽  
Gaussian Filter ◽  
Computationally Efficient ◽  
Remote Sensing Images ◽  
Gaussian Form ◽  
End To End ◽  
High Level

Remote sensing images are usually contaminated by cloud and corresponding shadow regions, making cloud and shadow detection one of the essential prerequisites for processing and translation of remote sensing images. Edge-precise cloud and shadow segmentation remains challenging due to the inherent high-level semantic acquisition of current neural segmentation fashions. We, therefore, introduce the Refined UNet series to partially achieve edge-precise cloud and shadow detection, including two-stage Refined UNet, v2 with a potentially efficient gray-scale guided Gaussian filter-based CRF, and v3 with an efficient multi-channel guided Gaussian filter-based CRF. However, it is visually demonstrated that the locally linear kernel used in v2 and v3 is not sufficiently sensitive to potential edges in comparison with Refined UNet. Accordingly, we turn back to the investigation of an end-to-end UNet-CRF architecture with a Gaussian-form bilateral kernel and its relatively efficient approximation. In this paper, we present Refined UNet v4, an end-to-end edge-precise segmentation network for cloud and shadow detection, which is capable of retrieving regions of interest with relatively tight edges and potential shadow regions with ambiguous edges. Specifically, we inherit the UNet-CRF architecture exploited in the Refined UNet series, which concatenates a UNet backbone of coarsely locating cloud and shadow regions and an embedded CRF layer of refining edges. In particular, the bilateral grid-based approximation to the Gaussian-form bilateral kernel is applied to the bilateral message-passing step, in order to ensure the delineation of sufficiently tight edges and the retrieval of shadow regions with ambiguous edges. Our TensorFlow implementation of the bilateral approximation is relatively computationally efficient in comparison with Refined UNet, attributed to the straightforward GPU acceleration. Extensive experiments on Landsat 8 OLI dataset illustrate that our v4 can achieve edge-precise cloud and shadow segmentation and improve the retrieval of shadow regions, and also confirm its computational efficiency.

Semantic Relation Model and Dataset for Remote Sensing Scene Understanding

2021 ◽  
Vol 10 (7) ◽  
pp. 488
Author(s):  
Peng Li ◽  
Dezheng Zhang ◽  
Aziguli Wulamu ◽  
Xin Liu ◽  
Peng Chen
Keyword(s):  
Remote Sensing ◽  
Scene Understanding ◽  
Deep Understanding ◽  
Remote Sensing Images ◽  
Convolutional Network ◽  
Scene Graph ◽  
Multi Scale ◽  
Relationship Extraction ◽  
High Level ◽  
Graph Generation

A deep understanding of our visual world is more than an isolated perception on a series of objects, and the relationships between them also contain rich semantic information. Especially for those satellite remote sensing images, the span is so large that the various objects are always of different sizes and complex spatial compositions. Therefore, the recognition of semantic relations is conducive to strengthen the understanding of remote sensing scenes. In this paper, we propose a novel multi-scale semantic fusion network (MSFN). In this framework, dilated convolution is introduced into a graph convolutional network (GCN) based on an attentional mechanism to fuse and refine multi-scale semantic context, which is crucial to strengthen the cognitive ability of our model Besides, based on the mapping between visual features and semantic embeddings, we design a sparse relationship extraction module to remove meaningless connections among entities and improve the efficiency of scene graph generation. Meanwhile, to further promote the research of scene understanding in remote sensing field, this paper also proposes a remote sensing scene graph dataset (RSSGD). We carry out extensive experiments and the results show that our model significantly outperforms previous methods on scene graph generation. In addition, RSSGD effectively bridges the huge semantic gap between low-level perception and high-level cognition of remote sensing images.

scholarly journals Shadow Detection and Compensation from Remote Sensing Images under Complex Urban Conditions

2021 ◽  
Vol 13 (4) ◽  
pp. 699
Author(s):  
Tingting Zhou ◽  
Haoyang Fu ◽  
Chenglin Sun ◽  
Shenghan Wang
Keyword(s):  
Remote Sensing ◽  
Urban Areas ◽  
Mean Shift ◽  
Color Space ◽  
Compensation Method ◽  
Shadow Detection ◽  
Processing Unit ◽  
Shadow Region ◽  
Remote Sensing Images ◽  
Shadow Compensation

Due to the block of high-rise objects and the influence of the sun’s altitude and azimuth, shadows are inevitably formed in remote sensing images particularly in urban areas, which causes missing information in the shadow region. In this paper, we propose a new method for shadow detection and compensation through objected-based strategy. For shadow detection, the shadow was highlighted by an improved shadow index (ISI) combined color space with an NIR band, then ISI was reconstructed by the objects acquired from the mean-shift algorithm to weaken noise interference and improve integrity. Finally, threshold segmentation was applied to obtain the shadow mask. For shadow compensation, the objects from segmentation were treated as a minimum processing unit. The adjacent objects are likely to have the same ambient light intensity, based on which we put forward a shadow compensation method which always compensates shadow objects with their adjacent non-shadow objects. Furthermore, we presented a dynamic penumbra compensation method (DPCM) to define the penumbra scope and accurately remove the penumbra. Finally, the proposed methods were compared with the stated-of-art shadow indexes, shadow compensation method and penumbra compensation methods. The experiments show that the proposed method can accurately detect shadow from urban high-resolution remote sensing images with a complex background and can effectively compensate the information in the shadow region.

scholarly journals The use of remote sensing satellite using deep learning in emergency monitoring of high-level landslides disaster in Jinsha River

2021 ◽  
Author(s):  
Leijin Long ◽  
Feng He ◽  
Hongjiang Liu
Keyword(s):  
Remote Sensing ◽  
Southwest China ◽  
Influence Factors ◽  
Classification Error ◽  
Model Parameters ◽  
Detection Accuracy ◽  
Remote Sensing Images ◽  
Jinsha River ◽  
Detection Model ◽  
High Level

AbstractIn order to monitor the high-level landslides frequently occurring in Jinsha River area of Southwest China, and protect the lives and property safety of people in mountainous areas, the data of satellite remote sensing images are combined with various factors inducing landslides and transformed into landslide influence factors, which provides data basis for the establishment of landslide detection model. Then, based on the deep belief networks (DBN) and convolutional neural network (CNN) algorithm, two landslide detection models DBN and convolutional neural-deep belief network (CDN) are established to monitor the high-level landslide in Jinsha River. The influence of the model parameters on the landslide detection results is analyzed, and the accuracy of DBN and CDN models in dealing with actual landslide problems is compared. The results show that when the number of neurons in the DBN is 100, the overall error is the minimum, and when the number of learning layers is 3, the classification error is the minimum. The detection accuracy of DBN and CDN is 97.56% and 97.63%, respectively, which indicates that both DBN and CDN models are feasible in dealing with landslides from remote sensing images. This exploration provides a reference for the study of high-level landslide disasters in Jinsha River.

Computationally Efficient Mean-Shift Parallel Segmentation Algorithm for High-Resolution Remote Sensing Images

2018 ◽  
Vol 46 (11) ◽  
pp. 1805-1814
Author(s):  
Tianjun Wu ◽  
Liegang Xia ◽  
Jiancheng Luo ◽  
Xiaocheng Zhou ◽  
Xiaodong Hu ◽  
...  
Keyword(s):  
Remote Sensing ◽  
High Resolution ◽  
Mean Shift ◽  
Segmentation Algorithm ◽  
Computationally Efficient ◽  
Remote Sensing Images

scholarly journals Coupling threshold theory and satellite-derived channel width to estimate the formative discharge of Himalayan foreland rivers

2021 ◽  
Vol 9 (1) ◽  
pp. 47-70
Author(s):  
Kumar Gaurav ◽  
François Métivier ◽  
Rajiv Sinha ◽  
Amit Kumar ◽  
Sampat Kumar Tandon ◽  
...  
Keyword(s):  
Remote Sensing ◽  
Satellite Images ◽  
Force Balance ◽  
Channel Width ◽  
Landsat 8 ◽  
Remote Sensing Images ◽  
Physical Force ◽  
Regulated Rivers ◽  
Threshold Theory ◽  
Discharge Regime

Abstract. We propose an innovative methodology to estimate the formative discharge of alluvial rivers from remote sensing images. This procedure involves automatic extraction of the width of a channel from Landsat Thematic Mapper, Landsat 8, and Sentinel-1 satellite images. We translate the channel width extracted from satellite images to discharge using a width–discharge regime curve established previously by us for the Himalayan rivers. This regime curve is based on the threshold theory, a simple physical force balance that explains the first-order geometry of alluvial channels. Using this procedure, we estimate the formative discharge of six major rivers of the Himalayan foreland: the Brahmaputra, Chenab, Ganga, Indus, Kosi, and Teesta rivers. Except highly regulated rivers (Indus and Chenab), our estimates of the discharge from satellite images can be compared with the mean annual discharge obtained from historical records of gauging stations. We have shown that this procedure applies both to braided and single-thread rivers over a large territory. Furthermore, our methodology to estimate discharge from remote sensing images does not rely on continuous ground calibration.

scholarly journals Potential Impact of Flood on Schistosomiasis in Poyang Lake Regions Based on Multi-Source Remote Sensing Images

2020 ◽  
Author(s):  
Jing-Bo Xue ◽  
Xin-Yi Wang ◽  
Li-Juan Zhang ◽  
Yu-Wan Hao ◽  
Zhe Chen ◽  
...  
Keyword(s):  
Remote Sensing ◽  
Water Body ◽  
Poyang Lake ◽  
Transmission Risk ◽  
Schistosomiasis Japonica ◽  
Landsat 8 ◽  
Remote Sensing Images ◽  
Poyang Lake Region ◽  
Lake Region ◽  
Dual Polarized

Abstract BackgroundFlooding may be the most important factors contributing to the rebound of Oncomelania hupensis in endemic foci. This study aimed to assess the risk of schistosomiasis japonica transmission impacted by flooding around the Poyang Lake region using multi-source remote sensing images.MethodsNormalized Difference Vegetation Index (NDVI) data collected by the Landsat 8 satellite was used as an ecological and geographical suitability indicator of O. hupensis snail habitats in the Poyang Lake region. The flood-affected water body expansion was estimated using dual polarized threshold calculations based on the dual polarized synthetic aperture radar (SAR). The image data were captured from Sentinel-1B satellite in May 2020 before the flood and in July 2020 during the flood. The spatial database of snail habitats distribution was created by using the 2016 snail survey in Jiangxi Province. The potential spread of O. hupensis snails after the flood was predicted by an overlay analysis of the NDVI maps of flood-affected water body areas. In addition, the risk of schistosomiasis transmission was classified based on O. hupensis snail density data and the related NDVI. ResultsThe surface area of Poyang Lake was approximately 2,207 km2 in May 2020 before the flood and 4,403 km2 in July 2020 during the period of the flood peak, and the flood-caused expansion of water body was estimated as 99.5%. After the flood, the potential snail habitats were predicted to be concentrated in areas neighboring the existing habitats in marshlands of the Poyang Lake. The areas with high risk of schistosomiasis transmission were predicted to be mainly distributed in Yongxiu, Xinjian, Yugan and Poyang (District) along Poyang Lake. By comparing the predictive results and actual snail distribution, the predictive accuracy of the model was estimated as 87%, which meant the 87% of actual snail distribution were correctly identified as the snail habitats in the model predictions. ConclusionsFlood-affected water body expansion and environmental factors pertaining to snail breeding may be rapidly extracted from Landsat 8 and Sentinel-1B remote sensing images. The applications of multi-source remote sensing data are feasible for the timely and effective assessment of the potential schistosomiasis transmission risk caused by snail spread during the flood disaster, which is of great significance for precision control of schistosomiasis.

Sign in / Sign up

Export Citation Format

Share Document