A GPU-Parallel Image Coregistration Algorithm for InSar Processing at the Edge

Romano, Diego; Lapegna, Marco

doi:10.3390/s21175916

Cited by 11 publications

(3 citation statements)

References 29 publications

(36 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To deal with the amount of data and intensive computing associated with InSAR missions, researchers have focused on performance optimization of InSAR processing platforms in recent years. Some of them have developed multi-core parallel algorithms [25][26][27] and GPU-based InSAR processing technology [28][29][30][31][32] to improve the overall performance; however, the potential of parallelism is strongly limited by the platform hardware. Meanwhile, other researchers have introduced InSAR HPC environments based on supercomputing platforms [33][34][35][36] or cloud computing platforms [37][38][39][40][41], which can utilize more computing power to realize large-scale or even national-scale remote sensing tasks.…”

Section: Introductionmentioning

confidence: 99%

A Parallel Sequential SBAS Processing Framework Based on Hadoop Distributed Computing

Wu,

Lv,

Yun

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

With the rapid development of microwave remote sensing and SAR satellite systems, the use of InSAR techniques has been greatly encouraged due to the abundance of SAR data with unprecedented temporal and spatial coverage. Small Baseline Subset (SBAS) is a promising time-series InSAR method for applications involving deformation monitoring of the Earth’s crust, and the sequential SBAS method is an extension of SBAS that allows long-term and large-scale surface displacements to be obtained with continuously auto-updating measurement results. As the Chinese LuTan-1 SAR system has begun acquiring massive SAR image data, the need for an efficient and lightweight InSAR processing platform has become urgent in various research fields. However, traditional sequential algorithms are incapable of meeting the huge challenges of low efficiency and frequent human interaction in large-scale InSAR data processing. Therefore, this study proposes a distributed parallel sequential SBAS (P2SBAS) processing chain based on Hadoop by effectively parallelizing and improving the current sequential SBAS method. P2SBAS mainly consists of two components: (1) a distributed SAR data storage platform based on HDFS, which supports efficient inter-node data transfer and continuous online data acquisition, and (2) several parallel InSAR processing algorithms based on the MapReduce model, including image registration, filtering, phase unwrapping, sequential SBAS processing, and so on. By leveraging the capabilities associated with the distributed nature of the Hadoop platform, these algorithms are able to efficiently utilize the segmentation strategy and perform careful boundary processing. These parallelized InSAR algorithm modules can achieve their goals on different nodes in the Hadoop distributed environment, thereby maximizing computing resources and improving the overall performance while comprehensively considering performance and precision. In addition, P2SBAS provides better computing and storage capabilities for small- and medium-sized teams compared to popular InSAR processing approaches based on cloud computing or supercomputing platforms, and it can be easily deployed on clusters thanks to the integration of various existing computing components. Finally, to demonstrate and evaluate the efficiency and accuracy of P2SBAS, we conducted comparative experiments on a set of 32 TerraSAR images of Beijing, China. The results demonstrate that P2SBAS can fully utilize various computing nodes to improve InSAR processing and can be applied well in large-scale LuTan-1 InSAR applications in the future.

show abstract

Section: Introductionmentioning

confidence: 99%

A Parallel Sequential SBAS Processing Framework Based on Hadoop Distributed Computing

Wu,

Lv,

Yun

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…The one-stage method is suitable for deployment on the terminal due to it only needing to be fed into the network once to predict all the bounding boxes, which is faster than the two-stage method. Many scholars focus on optimizing the speed of the one-stage SOTA algorithm to take advantage of its fast speed [23][24][25][26][27][28][29][30]. For example, in 2020, the article [23] proposed a mixed YOLOv3-LITE detector, this method complements Residual Blocks (ResBlocks) and parallel high-to-low resolution sub-networks, makes full use of shallow network features while increasing network depth, and uses "shallow and narrow" convolutional layers to achieve lightweight characteristic; in 2022, the article [24] proposed a YOLOv4-tiny based lightweight object detection framework to improves inference speed without sacrificing accuracy than baseline method.…”

Section: Introductionmentioning

confidence: 99%

“…Many scholars also focus on researching lightweight SAR image detection algorithms for high speed and low power consumption. In 2021, the article [25] proposed an efficient GPU par-allel algorithm to accelerate image registration for InSAR image processing and achieved 10w power consumption on Nvidia Jetson. In 2021, article [26] proposed a lightweight detection framework that integrates CFAR and YOLOv4 to achieve on-orbit target detection of SAR ship images.…”

Section: Introductionmentioning

confidence: 99%

Using Clean Energy Satellites to Interpret Imagery: A Satellite IoT Oriented Lightweight Object Detection Framework for SAR Ship Detection

Xie

Luo

et al. 2022

Sustainability

View full text Add to dashboard Cite

This paper studies the lightweight deep learning object detection algorithm to detect ship targets in SAR images that can be deployed on-orbit and accessed in the space-based IoT. Traditionally, remote sensing data must be transferred to the ground for processing. With the vigorous development of the commercial aerospace industry, computing, and high-speed laser inter-satellite link technologies, the interconnection of everything in the intelligent world has become an irreversible trend. Satellite remote sensing has entered the era of a big data link with IoT. On-orbit interpretation gives remote sensing images expanse application space. However, implementing on-orbit high-performance computing (HPC) is difficult; it is limited by the power and computer resource consumption of the satellite platform. Facing this challenge, building a processing algorithm with less computational complexity, less parameter quantity, high precision, and low computational power consumption is a key issue. In this paper, we propose a lightweight end-to-end SAR ship detector fused with the vision transformer encoder: YOLO−ViTSS. The experiment shows that YOLO−ViTSS has lightweight features, the model size is only 1.31 MB; it has anti-noise capability is suitable for processing SAR remote sensing images with native noise, and it also has high performance and low training energy consumption with 96.6 mAP on the SSDD dataset. These characteristics make YOLO−ViTSS suitable for porting to satellites for on-orbit processing and online learning. Furthermore, the ideas proposed in this paper help to build a cleaner and a more efficient new paradigm for remote sensing image interpretation. Migrating HPC tasks performed on the ground to on-orbit satellites and using solar energy to complete computing tasks is a more environmentally friendly option. This environmental advantage will gradually increase with the current construction of large-scale satellite constellations. The scheme proposed in this paper helps to build a novel real-time, eco-friendly, and sustainable SAR image interpretation mode.

show abstract