Orthomosaicking Thermal Drone Images of Forests via Simultaneously Acquired RGB Images
Bibliographic record
Abstract
Operational forest monitoring often requires fine-detail information in the form of an orthomosaic, created by stitching overlapping nadir images captured by aerial platforms such as drones. RGB drone sensors are commonly used for low-cost, high-resolution imaging that is conducive to effective orthomosaicking, but only capture visible light. Thermal sensors, on the other hand, capture long-wave infrared radiation, which is useful for early pest detection among other applications. However, these lower-resolution images suffer from reduced contrast and lack of descriptive features for successful orthomosaicking, leading to gaps or swirling artifacts in the orthomosaic. To tackle this, we propose a thermal orthomosaicking workflow that leverages simultaneously acquired RGB images. The latter are used for producing a surface mesh via structure from motion, while thermal images are only used to texture this mesh and yield a thermal orthomosaic. Prior to texturing, RGB-thermal image pairs are co-registered using an affine transformation derived from a machine learning technique. On average, the individual RGB and thermal images achieve a mutual information of 0.2787 after co-registration using our technique, compared to 0.0591 before co-registration, and 0.1934 using manual co-registration. We show that the thermal orthomosaic generated from our workflow (1) is of better quality than other existing methods, (2) is geometrically aligned with the RGB orthomosaic, (3) preserves radiometric information (i.e., surface temperatures) from the original thermal imagery, and (4) enables easy transfer of downstream tasks—such as tree crown detection from the RGB to the thermal orthomosaic. We also provide an open-source tool that implements our workflow to facilitate usage and further development.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".