Hyperspectral Quality Control
Once the hyperspectral processing is completed, it is crucial to examine the resulting reflectance to verify the data quality.
Strategies for quality control of hyperspectral data are provided here. The examples are performed using ENVI.
Reflectance on Targets
The reflectance on the targets should match the field measurements.
To verify the reflectance on the targets:
Define a Region of Interest (ROI) on each target
Evaluate the statistics of the reflectance within the ROI
Compare the average reflectance to the field measurements
Example
The figure below shows the reflectance statistics within the ROIs on three targets: A, B, and C. As evident in the figure, the reflectance remains stable after the 450nm wavelength, showing values of 0.54, 0.29, and 0.12 for Targets A, B, and C, respectively. These values align with the field measurements, which consistently read 0.53, 0.29, and 0.12 across all bands for Targets A, B, and C. Consequently, the results signify good performance of the ELM.
Addressing Suboptimal Results
The agreement between the target’s reflectance and field measurement is a key indicator of ELM performance. When faced with less-than-optimal results, consider employing the following strategies:
Refine target selection (refer to tips and examples here).
Use targets from a different cube or multiple cubes in target selection.
Last updated