Peer-reviewed research.
RASID’s work is grounded in published research. Team members co-author peer-reviewed papers on Earth observation, remote sensing, geospatial AI and quantum SAR processing with international collaborators. Further results, including the MethaneMapper validation work, are in preparation.
- 012026Earth Observation and Geomatics Engineering 10(1), 83–92journal article
Quantum Meets SAR: A Novel Range-Doppler Algorithm for Next-Gen Earth Observation
Khalil Al Salahat, Mohamad El Moussawi, Ali J. Ghandour
Proposes a Quantum Range-Doppler Algorithm that replaces the Fast Fourier Transform with the Quantum Fourier Transform to speed up the processing of raw synthetic-aperture radar signals into imagery. It also introduces a quantum implementation of Range Cell Migration Correction, the step that realigns returned echoes so a target’s energy lands in a single range bin.
Abstract
Synthetic Aperture Radar (SAR) plays a vital role in remote sensing due to its ability to capture high-resolution images regardless of weather conditions or daylight. However, to transform the raw SAR signals into interpretable imagery, advanced data processing techniques are essential. A widely used technique for this purpose is the Range Doppler Algorithm (RDA), which takes advantage of Fast Fourier Transform (FFT) to convert signals into the frequency domain for further processing. However, the computational cost of this approach becomes significant when dealing with large datasets. This paper presents a Quantum Range Doppler Algorithm (QRDA) that utilizes the Quantum Fourier Transform (QFT) to accelerate processing compared to the classical FFT. Furthermore, it introduces a quantum implementation of the Range Cell Migration Correction (RCMC) in the Fourier domain, a critical step in the RDA pipeline that realigns the received echoes so that the energy from a target is concentrated in a single range bin across all azimuth positions. The performance of the quantum RCMC is evaluated and compared against its classical counterpart, demonstrating the potential of quantum computing in advanced SAR imaging.
- 0220252025 IEEE International Conference on Next-Gen Technologies of Artificial Intelligence and Geoscience Remote Sensing (EarthSense)conference paper
A Decade of Wheat Mapping for Lebanon
Hasan Wehbi, Hasan Nasrallah, Mohamad Hasan Zahweh, Zeinab Takach, Veera Ganesh Yalla, Ali J. Ghandour
Combines a Temporal Spatial Vision Transformer with parameter-efficient fine-tuning and a field-boundary delineation step to map Lebanon's wheat fields from satellite imagery, addressing the common failure where many small parcels get merged into one field. The result is a decade of field-level wheat maps that can be counted and compared year over year, not just a pixel mask.
Abstract
Wheat accounts for approximatly 20% of the world’s caloric intake making it a vital component of global food secuirty. Given this significance, mapping wheat fields plays a crucial role in enabling various stakeholders including policymakers, researchers, and agricultural organizations to make informed decisions regarding food security, supply chain management, and resource allocation. In this paper, we tackle the problem of accurately mapping wheat fields out of satellite images by enhancing our previous work on winter wheat segmentation by creating an improved pipeline for processing data as well as presenting a decade-long analysis of wheat mapping in Lebanon. We integrate a Temporal Spatial Vision Transformer (TSViT) with Parameter-Efficient Fine Tuning (PEFT) and a novel post-processing pipeline based on the FOW delineation framework. Our enhanced pipeline addresses key challenges encountered in the previous approach such as clustering of small agricultural parcels in a single large field and sparse training labels. By merging wheat segmentation with precise field boundary extraction, our method produces geometrically coherent and semantically rich maps enabling us to perfom in-depth analysis such as calculating the total number of fields and tracking fields areas year over year. Extensive evaluations demonstrate improved boundary delineation and field-level precision, establishing the framework’s potential in operational agricultural monitoring and historical trend analysis. This work lays the foundation for a range of critical studies and future advancements. Building on the accurate mapping of wheat fields, our approach provides a crucial step toward more sophisticated agricultural analyses. Future work can extend this methodology to improve yield estimation, crop monitoring and broader analysis of agricultural trends.
- 032025IGARSS 2025 - 2025 IEEE International Geoscience and Remote Sensing Symposiumconference paper
Efficient Adaptation for Remote Sensing Visual Grounding
Hasan Moughnieh, Mohamad Chalhoub, Hasan Nasrallah, Cristiano Nattero, Paolo Campanella, Giovanni Nico, Ali J. Ghandour
Shows that large vision-language foundation models can be taught to find objects in satellite and aerial imagery from a plain-text description by tuning only a small fraction of their weights (LoRA, BitFit, adapters) instead of retraining the whole model. Accuracy matches or surpasses the state of the art while significantly reducing compute cost, which is what makes text-driven search over imagery affordable in production.
Abstract
Foundation models have revolutionized artificial intelligence (AI), offering remarkable capabilities across multi-modal domains. Their ability to precisely locate objects in complex aerial and satellite images, using rich contextual information and detailed object descriptions, is essential for remote sensing (RS). These models can associate textual descriptions with object positions through the Visual Grounding (VG) task, but due to domain-specific challenges, their direct application to RS produces sub-optimal results. To address this, we applied Parameter Efficient Fine Tuning (PEFT) techniques to adapt these models for RS-specific VG tasks. Specifically, we evaluated LoRA placement across different modules in Grounding DINO and used BitFit and adapters to fine-tune the OFA foundation model pre-trained on general-purpose VG datasets. This approach achieved performance comparable to or surpassing current State Of The Art (SOTA) models while significantly reducing computational costs. This study highlights the potential of PEFT techniques to advance efficient and precise multi-modal analysis in RS, offering a practical and cost-effective alternative to full model training.