A Framework for a High-Performance Remote Sensing Image Real-Time Processing System
- Version
- Download 20
- File Size 373.95 KB
- File Count 1
- Create Date 12 November 2025
- Last Updated 12 November 2025
A Framework for a High-Performance Remote Sensing Image Real-Time Processing System
Lajja Dave 1, Jigar Dalvadi 2
1Sardar Patel College of Engineering and Technology, Bakrol, Anand
2 Sardar Patel College of Engineering and Technology, Bakrol, Anand
Abstract - High-performance real-time processing systems for remote sensing imaging are critically needed due to the exponential growth in data volume from satellite constellations and unmanned aerial vehicles (UAVs). For time-sensitive applications like disaster response, military surveillance, and environmental monitoring, traditional post-processing techniques—which entail downloading, storing, and then analyzing data—introduce substantial latency. The architectural elements and computational paradigms necessary to construct a high-performance real-time processing system are examined in this study. We provide an extensive review of the literature that traces the development of stream-based systems from batch processing. The fundamental techniques are divided into two categories: software algorithmic approaches (such as stream processing frameworks and lightweight deep learning models) and hardware acceleration strategies (such as GPUs, FPGAs, and specialized AI chips). For the majority of jobs, a hybrid CPU-GPU architecture that uses optimized convolutional neural networks (CNNs) and FPGA-based pre-processing provides the best performance, flexibility, and energy economy, according to a comparative analysis. The accuracy-speed trade-off in algorithms, system integration complexity, and data throughput constraints are some of the major issues that are discussed. We conclude that edge-computing paradigms and AI-driven, adaptive processing pipelines that may independently prioritize jobs based on operational needs are key to the future of real-time remote sensing.
Key Words: Real-time processing, GPU computing, Lightweight deep learning, Environmental monitoring, Time-sensitive applications, Low-latency systems, Accuracy-speed trade-off
Download