scholarly journals Semi-Supervised SAR Target Detection Based on an Improved Faster R-CNN

2021 ◽  
Vol 14 (1) ◽  
pp. 143
Author(s):  
Leiyao Liao ◽  
Lan Du ◽  
Yuchen Guo

In the remote sensing image processing field, the synthetic aperture radar (SAR) target-detection methods based on convolutional neural networks (CNNs) have gained remarkable performance relying on large-scale labeled data. However, it is hard to obtain many labeled SAR images. Semi-supervised learning is an effective way to address the issue of limited labels on SAR images because it uses unlabeled data. In this paper, we propose an improved faster regions with CNN features (R-CNN) method, with a decoding module and a domain-adaptation module called FDDA, for semi-supervised SAR target detection. In FDDA, the decoding module is adopted to reconstruct all the labeled and unlabeled samples. In this way, a large number of unlabeled SAR images can be utilized to help structure the latent space and learn the representative features of the SAR images, devoting attention to performance promotion. Moreover, the domain-adaptation module is further introduced to utilize the unlabeled SAR images to promote the discriminability of features with the assistance of the abundantly labeled optical remote sensing (ORS) images. Specifically, the transferable features between the ORS images and SAR images are learned to reduce the domain discrepancy via the mean embedding matching, and the knowledge of ORS images is transferred to the SAR images for target detection. Ultimately, the joint optimization of the detection loss, reconstruction, and domain adaptation constraints leads to the promising performance of the FDDA. The experimental results on the measured SAR image datasets and the ORS images dataset indicate that our method achieves superior SAR target detection performance with limited labeled SAR images.

Sensors ◽  
2019 ◽  
Vol 19 (23) ◽  
pp. 5284 ◽  
Author(s):  
Heng Zhang ◽  
Jiayu Wu ◽  
Yanli Liu ◽  
Jia Yu

In recent years, the research on optical remote sensing images has received greater and greater attention. Object detection, as one of the most challenging tasks in the area of remote sensing, has been remarkably promoted by convolutional neural network (CNN)-based methods like You Only Look Once (YOLO) and Faster R-CNN. However, due to the complexity of backgrounds and the distinctive object distribution, directly applying these general object detection methods to the remote sensing object detection usually renders poor performance. To tackle this problem, a highly efficient and robust framework based on YOLO is proposed. We devise and integrate VaryBlock to the architecture which effectively offsets some of the information loss caused by downsampling. In addition, some techniques are utilized to facilitate the performance and to avoid overfitting. Experimental results show that our proposed method can enormously improve the mean average precision by a large margin on the NWPU VHR-10 dataset.


2019 ◽  
Vol 11 (6) ◽  
pp. 631 ◽  
Author(s):  
Shaoming Zhang ◽  
Ruize Wu ◽  
Kunyuan Xu ◽  
Jianmei Wang ◽  
Weiwei Sun

Offshore and inland river ship detection has been studied on both synthetic aperture radar (SAR) and optical remote sensing imagery. However, the classic ship detection methods based on SAR images can cause a high false alarm ratio and be influenced by the sea surface model, especially on inland rivers and in offshore areas. The classic detection methods based on optical images do not perform well on small and gathering ships. This paper adopts the idea of deep networks and presents a fast regional-based convolutional neural network (R-CNN) method to detect ships from high-resolution remote sensing imagery. First, we choose GaoFen-2 optical remote sensing images with a resolution of 1 m and preprocess the images with a support vector machine (SVM) to divide the large detection area into small regions of interest (ROI) that may contain ships. Then, we apply ship detection algorithms based on a region-based convolutional neural network (R-CNN) on ROI images. To improve the detection result of small and gathering ships, we adopt an effective target detection framework, Faster-RCNN, and improve the structure of its original convolutional neural network (CNN), VGG16, by using multiresolution convolutional features and performing ROI pooling on a larger feature map in a region proposal network (RPN). Finally, we compare the most effective classic ship detection method, the deformable part model (DPM), another two widely used target detection frameworks, the single shot multibox detector (SSD) and YOLOv2, the original VGG16-based Faster-RCNN, and our improved Faster-RCNN. Experimental results show that our improved Faster-RCNN method achieves a higher recall and accuracy for small ships and gathering ships. Therefore, it provides a very effective method for offshore and inland river ship detection based on high-resolution remote sensing imagery.


2021 ◽  
Vol 13 (13) ◽  
pp. 2558
Author(s):  
Lei Yu ◽  
Haoyu Wu ◽  
Zhi Zhong ◽  
Liying Zheng ◽  
Qiuyue Deng ◽  
...  

Synthetic aperture radar (SAR) is an active earth observation system with a certain surface penetration capability and can be employed to observations all-day and all-weather. Ship detection using SAR is of great significance to maritime safety and port management. With the wide application of in-depth learning in ordinary images and good results, an increasing number of detection algorithms began entering the field of remote sensing images. SAR image has the characteristics of small targets, high noise, and sparse targets. Two-stage detection methods, such as faster regions with convolution neural network (Faster RCNN), have good results when applied to ship target detection based on the SAR graph, but their efficiency is low and their structure requires many computing resources, so they are not suitable for real-time detection. One-stage target detection methods, such as single shot multibox detector (SSD), make up for the shortage of the two-stage algorithm in speed but lack effective use of information from different layers, so it is not as good as the two-stage algorithm in small target detection. We propose the two-way convolution network (TWC-Net) based on a two-way convolution structure and use multiscale feature mapping to process SAR images. The two-way convolution module can effectively extract the feature from SAR images, and the multiscale mapping module can effectively process shallow and deep feature information. TWC-Net can avoid the loss of small target information during the feature extraction, while guaranteeing good perception of a large target by the deep feature map. We tested the performance of our proposed method using a common SAR ship dataset SSDD. The experimental results show that our proposed method has a higher recall rate and precision, and the F-Measure is 93.32%. It has smaller parameters and memory consumption than other methods and is superior to other methods.


Author(s):  
Tarik Benabdelouahab ◽  
Hayat Lionboui ◽  
Rachid Hadria ◽  
Riad Balaghi ◽  
Abdelghani Boudhar ◽  
...  

Irrigated agriculture is an important strategic sector for Morocco, contributing to food security and employment. Nowadays, irrigation scheme managers shall ensure that water is optimally used. The main objective was to support the irrigation monitoring and management of wheat in the irrigated perimeter using optical remote sensing and crop modeling. The potential of spectral indices derived from SPOT-5 images was explored for quantifying and mapping surface water content changes at large scale. Indices were computed using the reflectance in red, near infrared, and shortwave infrared bands. A field crop model (AquaCrop) was adjusted and tested to simulate the grain yield and the temporal evolution of soil moisture status. This research aimed at providing a scientific and technical approach to assist policymakers and stakeholders to improve monitoring irrigation and mitigating wheat water stress at field and irrigation perimeter levels in semi-arid areas. The approach could lead to operational management tools for an efficient irrigation at field and regional levels.


2017 ◽  
Vol 37 (10) ◽  
pp. 1011004 ◽  
Author(s):  
宋明珠 Song Mingzhu ◽  
曲宏松 Qu Hongsong ◽  
金 光 Jin Guang

Author(s):  
Changmiao Hu ◽  
Ping Tang

In recent years, China's demand for satellite remote sensing images increased. Thus, the country launched a series of satellites equipped with high-resolution sensors. The resolutions of these satellites range from 30 m to a few meters, and the spectral range covers the visible to the near-infrared band. These satellite images are mainly used for environmental monitoring, mapping, land surface classification and other fields. However, haze is an important factor that often affects image quality. Thus, dehazing technology is becoming a critical step in high-resolution remote sensing image processing. This paper presents a rapid algorithm for dehazing based on a semi-physical haze model. Large-scale median filtering technique is used to extract large areas of bright, low-frequency information from images to estimate the distribution and thickness of the haze. Four images from different satellites are used for experiment. Results show that the algorithm is valid, fast, and suitable for the rapid dehazing of numerous large-sized high-resolution remote sensing images in engineering applications.


2016 ◽  
Vol 76 (12) ◽  
pp. 14461-14483 ◽  
Author(s):  
Yudong Lin ◽  
Hongjie He ◽  
Heng-Ming Tai ◽  
Fan Chen ◽  
Zhongke Yin

2019 ◽  
Vol 11 (10) ◽  
pp. 1153 ◽  
Author(s):  
Mesay Belete Bejiga ◽  
Farid Melgani ◽  
Pietro Beraldini

Learning classification models require sufficiently labeled training samples, however, collecting labeled samples for every new problem is time-consuming and costly. An alternative approach is to transfer knowledge from one problem to another, which is called transfer learning. Domain adaptation (DA) is a type of transfer learning that aims to find a new latent space where the domain discrepancy between the source and the target domain is negligible. In this work, we propose an unsupervised DA technique called domain adversarial neural networks (DANNs), composed of a feature extractor, a class predictor, and domain classifier blocks, for large-scale land cover classification. Contrary to the traditional methods that perform representation and classifier learning in separate stages, DANNs combine them into a single stage, thereby learning a new representation of the input data that is both domain-invariant and discriminative. Once trained, the classifier of a DANN can be used to predict both source and target domain labels. Additionally, we also modify the domain classifier of a DANN to evaluate its suitability for multi-target domain adaptation problems. Experimental results obtained for both single and multiple target DA problems show that the proposed method provides a performance gain of up to 40%.


Data in Brief ◽  
2020 ◽  
Vol 33 ◽  
pp. 106362
Author(s):  
Sudhanshu Shekhar Jha ◽  
Manohar Kumar ◽  
Rama Rao Nidamanuri

Sign in / Sign up

Export Citation Format

Share Document