Figure 1 - uploaded by Zhi Zhao
Content may be subject to copyright.
Source publication
Ship surveillance is important for maritime security and safety. It plays important roles in many applications including ocean environment monitoring, search and rescue, anti-piracy and military reconnaissance. Among various sensors used for maritime surveillance, space-borne Synthetic Aperture Radar (SAR) is valued for its high resolution over wid...
Context in source publication
Similar publications
Ship detection and classification is critical for national maritime security and national defense. Although some SAR (Synthetic Aperture Radar) image-based ship detection approaches have been proposed and used, they are not able to satisfy the requirement of real-world applications as the number of SAR sensors is limited, the resolution is low, and...
Synthetic aperture radar (SAR), in along-track interferometry mode, is extensively used in sensing oceanic surface. Detecting moving ships is becoming an increasingly important requirement in global monitoring of environment and maritime security. To realize this requirement, we first constructed a metric for detecting moving ships in SAR images by...
Citations
... In this trend, it is imperative to establish an effective maritime surveillance system (Bambulyak and Ehlers, 2020). Thus, countries facilitate Automatic Identification System (AIS) receivers, radar, and marine cameras for maritime surveillance (Zhao et al., 2014). Among the systems, AIS emerged as the most effective, as it provides real-time information on ships' identification, location, and speed according to internationally regularized rules. ...
... To mitigate the limitations associated with transmission distance, low-orbit satellite-based AIS has been adopted in several countries, such as the United States, Canada, Germany, Norway, Italy, and China (Zhao et al., 2014). However, accessing the AIS data continues to be challenging due to the exclusive costliness of commercial products, and the limited availability of government-owned products. ...
Synthetic Aperture Radar (SAR) presents a valuable capability for detecting objects over a large expanse of land or sea, independent of prevailing day and weather conditions. SAR can effectively identify targets by utilizing electromagnetic waves with high penetration rates through the atmosphere. The reflective nature of radar signals by ships that are often constructed with metal and complex superstructures makes satellite-based SAR an ideal tool for comprehensive maritime surveillance.
In recent times, advancements in microsatellites have led to satellite SAR products exhibiting high spatial and temporal resolutions. Consequently, the need for efficient ship detection methods has become increasingly apparent. One approach commonly employed in the past involves detecting ships from a single-channel SAR image, typically utilizing window-based algorithms or deep learning-based detection algorithms.
However, the window-based Constant False Alarm Rate (CFAR) algorithm is inefficient due to its requirement to examine each pixel individually. On the other hand, deep learning algorithms present their challenges, as they require large amounts of training data for model generation and involve high computational complexity. In order to overcome these limitations, this study introduces an innovative and efficient ship detection method that eliminates a training process and offers lightweight computation. This is achieved by synergistically combining the strengths of the CFAR threshold determination method with the candidate proposal of deep learning-based algorithms.
To accomplish this objective, this paper proposes a three-stage ship detection process, encompassing the generation of tile images, candidate extraction through prescreening, and subsequent discrimination of the candidates. By adopting a stepwise approach that specifically aims the reduce false detections, the proposed method exhibits high efficiency, as only the identified candidates undergo the discrimination process.
To enrich the analysis and validate the detected ships, marine traffic data in the form of Automatic Identification System (AIS) and Vessel Pass (V-Pass) information are utilized. Additionally, the state vector estimation method proposed by previous researchers is employed to approximate the actual positions of the satellite and ships at the time of detection. Moreover, a clustering method is ingeniously applied to refine the AIS data, considering the inherent mixed trajectory problem associated with AIS information.
The detection rate is evaluated for ship detection results using a dataset comprising 46 and 52 Sentinel-1 images in Busan and Incheon, and corresponding marine traffic data collected, spanning the year 2018.
In Busan waters, the proposed ship detection method successfully identifies 84.09% of the ships that have available length information in the AIS on an annual basis. On the other hand, it detects 43.73% of the ships that lack length information in the AIS. Additionally, the method is capable of detecting 11.13% of the V-Pass ships. Similarly, the ship detection in Incheon achieves a detection rate of approximately 79.99% for ships with length information in the AIS, while successfully identifying 43.14% of ships without length information in the AIS. On the other hand, it detects 5.91% of the V-Pass ships in the region. The reason for the difference in detection rate along with data type is presented through the analysis of the relationship between ship characteristics and significant wave height.
The effectiveness of the proposed method was assessed through a comparative analysis with previously established CFAR models, such as CA-CFAR and MAD-CFAR. The prescreen duration of the proposed method was found to be approximately 74 times shorter than that of the CA-CFAR model and approximately 149 times shorter than that of the MAD-CFAR model. In addition, no significant differences in detection rate and false alarm rate are found in statistical tests.
The spatial locations of the detected ships are leveraged to assess the wider marine traffic density across extended areas, considering the transmission distance limitations of AIS and V-Pass systems. This assessment aims to explore the potential use of density data derived from the ship detection results in contributing to marine traffic assessment.
... One of the most important ventures that has gained much attention from many countries is the use of data fusion (Butler, 2005;Katsilieris et al., 2013;Kurekin et al., 2019;Li et al., 2021;Longépé et al., 2018;Zhao et al., 2014). Many countries and authorities have ventured into integrating data sources for monitoring activities in the maritime domain in all possible aspects. ...
Uninterrupted monitoring of illegal activities occurring in a large maritime area with numerous vessels is very challenging. This article reports the development of system for automated monitoring of navigational data obtained via the automatic identification system (AIS). The AIS information obtained from a vessel is used to build up a time series dataset where an autoregression integrated moving average (ARIMA) model is used on the dataset to predict the status and future position of the vessel. Since the actual navigational trajectories of vessels are predictable, the projected information obtained from the ARIMA model can be compared against the next actual AIS information. The model could decide to trigger a warning alert using preset criteria after comparing the prediction to the actual data. This article shows two cases where a particular ship displayed suspicious behaviours, prompting the model to trigger a warning. While the preset criteria can initially be decided by the user, such criteria can be shaped or trained ‘on the fly’ to produce more accurate decisions as more similar cases are detected. The author believes that the ARIMA model is simple and robust for monitoring suspicious behaviours. It is versatile and warning criteria can be defined and shaped according to user requirements.
... Ship detection has received increasing attention in the field of synthetic aperture radar (SAR) for its broad application for military [1][2][3], marine traffic monitoring [4,5], harbor surveillance [6,7], etc. Following their success in optical images, deep learningbased methods are replacing traditional methods, which rely on manually designed feature extraction, as the most popular models for SAR images [8][9][10][11][12][13]. ...
Although they have achieved great success in optical images, deep convolutional neural networks underperform for ship detection in SAR images because of the lack of color and textual features. In this paper, we propose our framework which integrates prior knowledge into neural networks by means of the attention mechanism. Because the background of ships is mostly water surface or coast, we use clustering algorithms to generate the prior knowledge map from brightness and density features. The prior knowledge map is later resized and fused with convolutional feature maps by the attention mechanism. Our experiments demonstrate that our framework is able to improve various one-stage and two-stage object detection algorithms (Faster R-CNN, RetinaNet, SSD, and YOLOv4) on two benchmark datasets (SSDD, LS-SSDD, and HRSID).
... In contrast to CFAR detectors and its variations, CNN analyzed the target pattern and its pixel arrangement utilizing repetitive convolution calculator [7]. Given that the vessel backscattering signature in SAR was presented as a summation of different targets in each pixel [8], application of CNN-based detector in SAR image was able to be regarded effective in aspect of robustness, especially when discriminating vessel-like targets in coastal regions: bridges, buoys and small islands [9]. ...
... The notation denotes the speed of light, while signifies range frequency deviation. Incorporating (3) and (8), the quadratic phase function in range frequency-azimuth time domain is derived as in (9). ...
... A number of studies on vessel monitoring focused on performance enhancement from modifying and comparing different detection algorithms [9], [10], [14], [15]. When it came to utilizing detectors based on artificial intelligence however, quantity and quality of training data were decisive in their accuracy [28]. ...
Along with vessel detection, vessel recognition in high-resolution SAR images was necessary in order to monitor marine vessels effectively. However, lack of target data and phase defocusing of target from its velocity limited the recognition performance, especially when using detectors based on artificial intelligence. This study accordingly proposed effective vessel recognition in high-resolution ICEYE spotlight SAR images consecutively utilizing (i) vessel detector robust to defocused moving vessels and (ii) mitigation of moving target phase distortion. In order to apply quantitative and qualitative training data enhancement, a target velocity SAR phase refocusing function was developed. The proposed target velocity SAR phase refocusing function generated defocused SLC image with respect to different target azimuth velocity, which can be utilized for both training data augmentation and refocusing of velocity-induced phase distortion. Achievement of stable vessel recognition performance was enabled from (i) robust vessel detection on defocused moving vessels and (ii) well-focused detected vessel targets, both of which were consecutively applied using the proposed target velocity SAR phase refocusing function. Vessel detection results demonstrated robust performance regardless of vessel motion and vessel recognition results significantly improved after phase refocusing, both of which were subject to quantitative and qualitative training data enhancement. Performance of the proposed algorithm was analyzed both in terms of phase focusing and velocity estimation. Refocusing performance outperformed that of conventional state-of-the-art autofocusing algorithm, modified Phase Gradient Autofocusing, while azimuth velocity estimation derived the average offset of 0.68 m/s, which was regarded more accurate than previous azimuth velocity estimators based on single-channel SAR image.
... There exists a wealth of spatiotemporal data-driven solutions related to these issues that build upon a multitude of features from vessel tracking devices and structural properties of ships, weather conditions, and internal machinery sensors. Tracking can be performed using synthetic aperture radar (SAR) images and data from an automatic identification system (AIS) [1] or other surveillance systems [2]. The spatial dimensions focus on local conditions, e.g., wave-energy spectra and currents, which affect the costs of the overseas movement of a ship, while the temporal aspects correspond to environmental and ship-system conditions, and examine how they evolve with time (e.g., speed, power absorbed by the ship propulsion system, rotational motions along the three axes of the vessel). ...
The operation of any vessel includes risks, such as mechanical failure, collision, property loss, cargo loss, or damage. For modern container ships, safe navigation is challenging as the rate of innovation regarding design, speed profiles, and carrying capacity has experienced exponential growth over the past few years. Prevention of cargo loss in container ship liners is of high importance for the Maritime industry and the waterborne sector as it can lead to potentially disastrous, harmful, or even life-threatening outcomes for the crew, the shipping company, the marine environment, and aqua-culture. With the installment of onboard decision support system(s) (DSS) that will provide the required operational guidance to the vessel’s master, we aim to prevent and overcome such events. This paper explores cargo losses in container ships by employing a novel weather routing optimization DS framework that aims to identify excessive motions and accelerations caused by bad weather at specific times and locations; it also suggests alternative routes and, thus, ultimately prevents cargo loss and damage.
... It has the advantages of all-weather and all-time working performance. Accordingly, they are widely applied in the marine field, such as ship surveillance [2], marine environmental protection [3] shipwreck rescue [4], and sea ice classification [5]. Currently, a major current focus of SAR in the marine field is ship detection in images [6], which makes it possible to use the deep learning method for SAR ship detection. ...
In the field of ship detection, most research on lightweight models comes at the expense of accuracy. This study aims to address this challenge through a deep learning approach and proposes a model DWSC-YOLO, which is inspired by YOLOv5 and MobileNetV3. The model employs a lightweight framework as the backbone network, and the activation function and attention mechanism are researched. Furthermore, to improve the accuracy of the convolutional neural network and reduce loss, heterogeneous convolutions are added to the network. Three independent experiments were carried out using the proposed model. The experiment results show that the model can achieve excellent detection results with a small number of computational resources and costs. The mAP of the model is 99.5%, the same as YOLOv5, but the volume is 2.37 M, which is 79.8% less.
... Finally, Section 9 concludes the paper. [36] 2022 pedestrians/cars tracking UAV data Siamese networks [5] 2020 wildfire observation UAV data traditional technique [37] 2021 fire detection and analysis satellite multi-spectral data traditional technique [38] 2014 Ship Surveillance space-borne SAR and AIS traditional technique ...
Video satellites have recently become an attractive method of Earth observation, providing consecutive images of the Earth’s surface for continuous monitoring of specific events. The development of on-board optical and communication systems has enabled the various applications of satellite image sequences. However, satellite video-based target tracking is a challenging research topic in remote sensing due to its relatively low spatial and temporal resolution. Thus, this survey systematically investigates current satellite video-based tracking approaches and benchmark datasets, focusing on five typical tracking applications: traffic target tracking, ship tracking, typhoon tracking, fire tracking, and ice motion tracking. The essential aspects of each tracking target are summarized, such as the tracking architecture, the fundamental characteristics, primary motivations, and contributions. Furthermore, popular visual tracking benchmarks and their respective properties are discussed. Finally, a revised multi-level dataset based on wpafb videos is generated and quantitatively evaluated for future development in the satellite video-based tracking area. In addition, 54.3% of the tracklets with lower ds are selected and renamed as the Easy group, while 27.2% and 18.5% of the tracklets are grouped into the Medium-ds group and the Hard-ds group, respectively.
... Ship detection in high-resolution SAR images has attracted much attention due to its broad application prospects. Many traditional methods [8][9][10] have been proposed to detect multi-scale ships in complex environments. For example, [9,10] separated the land from the sea and then detected the object, and identified the object based on artificial features. ...
... Many traditional methods [8][9][10] have been proposed to detect multi-scale ships in complex environments. For example, [9,10] separated the land from the sea and then detected the object, and identified the object based on artificial features. Wackerman et al. [8] proposed a twoparameter constant false-alarm rate (CFAR) algorithm, which can adaptively adjust the threshold and use the estimated statistical distribution to distinguish objects from the background through the computed threshold. ...
Ship detection in Synthetic Aperture Radar (SAR) is a challenging task due to the random orientation of the ship and discrete appearance caused by radar signal. In this paper, We introduce a novel unsupervised domain adaptation framework for ship detection in SAR images by employing context-preserving region-based contrastive learning. We enhance the ship detection in SAR by learning knowledge from both labeled remote sensing optical image domain and unlabeled SAR image domain. Additionally, we propose a pseudo feature generation network to generate pseudo domain samples for augmenting pseudo-features. Specifically, we refine the pseudo-features by calculating a region-based contrastive loss on the features extracted from the object region and the background region to capture the contextual information for SAR ship detection. Extensive experiments and visualizations show that our method can outperform the state-of-the-art and have good generalization performance.
... Although traditional methods have achieved good results, there are some challenges. For example, traditional methods (Chen et al. 2017;Zhi et al. 2014;Fingas and Brown 2014) detect after sea-land segmentation and utilize the hand-crafted features for discrimination. But they have poor performance in near-shore areas and have difficulty ruling out false alarms. ...
Ship detection is significant for monitoring ports, especially contributing to the safe driving of Unmanned Surface Vehicle (USV). However, recent ship detection based on deep learning lacks complete ship datasets and uses the classification score as the ranking basis, which harms their performance. To address the problems, we present a one-stage localization estimation detector (LEDet) with ship-customized data augmentation. Specifically, we integrate the localization quality estimation into the classification branch as a soft label localization score. We further apply ship-customized data augmentation named “cutting-transform-paste” to expand ship datasets without manual annotation. Hence, a large number of diverse ship datasets can be created. Extensive experiments show that our LEDet consistently exceeds the strong baseline by 8.0% COCO-style Average Precision (AP) with ResNet-50. It significantly improves the ship detection performance.
... Other studies focus on retrieving orientation after detection or on refocusing moving targets (Jin et al., 2017). Integrated detection receives a lot of attention, with Automatic Identification System and SAR being complementary (Zhao et al., 2014). ...
In this review, we present the main approaches developed around satellite and airborne Synthetic Aperture Radar (SAR) imagery. The great range of SAR imagery applications is summarized in this paper. We organize the most popular methods and their applications in a cohesive manner. SAR data applications are classified into earth observation and object detection applications and the former are separated into land, sea, and ice applications. We present the basic methodologies and recent advances in land cover classification and object detection, as well as techniques for parameter retrieval from SAR data. We give advantages and disadvantages and highlight the particular characteristics of each method. It is shown that usage of SAR contributes to the amelioration of techniques and to the enhancement of reliability.
