Image analysis of hyperspectral and multispectral data using projection pursuit
Bibliographic record
Abstract
Given recent advancements of modern hyperspectral (HS) sensors, the potential for information extraction has increased drastically given the continual improvements in spatial and spectral resolution. As a result, more sophisticated feature extraction and target detection (TD) algorithms are needed to improve the performance of the image analyst, whether computer-based or human. In this paper, a novel TD algorithm based on Projection Pursuit (PP) is proposed and implemented. PP is a well-known technique for dimensionality reduction in multi-band data sets without loss of any critical information. This technique highlights different features of interest in an image, thus improving and simplifying subsequent anomaly detection. The new target detection technique is based on a hybrid of PP and Reed_Xiaoli (RX) anomaly detector. In this study, the combining of PP with the RX detector (PPRX) adds some extra value to the standard RX detection technique and leads the development of a TD method that can be applied on hyperspectral/multispectral (MS) data sets. This novel technique, after being trained by using the Projection Index (PI) and a priori information of target of interest, utilizes RX detector to evaluate each potential projection. The main drawback of previously introduced PP methods such as those based on Information Divergence and Kurtosis/Skewness is that these techniques are sensitive to statistical outliers and cannot be used to highlight a specific target of interest. This study uses three data sets: (1) 4-band IKONOS multispectral data (2) 210-band HYDICE, and (3) 200-band simulated hyperspectral data set.
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 machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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 source (direct Gemma or distilled Codex), 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".