Towards Efficient Mangrove Monitoring: Super-Resolution Trained on Paired Low- and High-Altitude UAV Images

Mangrove forests are carbon-rich, yet their dense canopies and intertidal setting hinder ground-based monitoring. Unmanned aerial vehicles (UAVs) offer an alternative but impose a trade-off: high-altitude flights cover larger areas, whereas low-altitude flights resolve individual crowns. This study proposes a field-based protocol for super-resolution of mangrove UAV imagery trained on image pairs acquired at two different flight altitudes. Images were acquired at 82 locations in the Ao Kung mangrove forest, Phuket, Thailand, at 80.3 m and 15.6 m to provide low-resolution inputs and high-resolution references, and 64 pairs were retained after registration. A very deep super-resolution (VDSR) network achieved a mean peak signal-to-noise ratio (PSNR) gain of 1.11 dB over bicubic interpolation on six held-out test pairs. In a five-fold spatially blocked cross-validation of all 64 pairs, VDSR outperformed bicubic interpolation on every pair, with a mean PSNR gain of 0.74 dB (95% CI [0.68, 0.80]). However, a brightness and contrast correction of the bicubic image, fitted on the training pairs, reached the same mean PSNR, so the gain of VDSR arose from radiometric correction rather than from recovered canopy detail. The protocol therefore provides real paired-altitude data on which super-resolution models can be trained and tested, and models that tolerate residual misalignment between the paired images should be evaluated in future work.

Authors

Institutions

Publication Details

Journal
Environments
Published
2026-11-18
DOI
https://doi.org/10.3390/environments13110631
Primary Topic
Advanced Image Processing Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Towards Efficient Mangrove Monitoring: Super-Resolution Trained on Paired Low- and High-Altitude UAV Images

Pawita Boonrat, Sunil Nepali, Veeranun Songsom
1 links
Environments
Advanced Image Processing Techniques
article

Towards Efficient Mangrove Monitoring: Super-Resolution Trained on Paired Low- and High-Altitude UAV Images

Pawita Boonrat, Sunil Nepali, Veeranun Songsom
article en

Abstract

Mangrove forests are carbon-rich, yet their dense canopies and intertidal setting hinder ground-based monitoring. Unmanned aerial vehicles (UAVs) offer an alternative but impose a trade-off: high-altitude flights cover larger areas, whereas low-altitude flights resolve individual crowns. This study proposes a field-based protocol for super-resolution of mangrove UAV imagery trained on image pairs acquired at two different flight altitudes. Images were acquired at 82 locations in the Ao Kung mangrove forest, Phuket, Thailand, at 80.3 m and 15.6 m to provide low-resolution inputs and high-resolution references, and 64 pairs were retained after registration. A very deep super-resolution (VDSR) network achieved a mean peak signal-to-noise ratio (PSNR) gain of 1.11 dB over bicubic interpolation on six held-out test pairs. In a five-fold spatially blocked cross-validation of all 64 pairs, VDSR outperformed bicubic interpolation on every pair, with a mean PSNR gain of 0.74 dB (95% CI [0.68, 0.80]). However, a brightness and contrast correction of the bicubic image, fitted on the training pairs, reached the same mean PSNR, so the gain of VDSR arose from radiometric correction rather than from recovered canopy detail. The protocol therefore provides real paired-altitude data on which super-resolution models can be trained and tested, and models that tolerate residual misalignment between the paired images should be evaluated in future work.

EnvironmentsVol. 13(11)
Prince of Songkla University (TH)
Life on land
Advanced Image Processing Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon