Development of a Processing Workflow for Hyperspectral Images
Presentation Type
Poster
Faculty Advisor
Stefan Robila
Access Type
Event
Start Date
26-4-2023 11:00 AM
End Date
26-4-2023 12:00 PM
Description
Hyperspectral Images (HSI) are collection of grayscale images of the same scene taken over narrow light wavelength intervals. HSI have been used extensively in various fields such as earth sciences, manufacturing, agriculture, food safety, defense, and homeland security. Hyperspectral data has become more accessible due to the increasing number of sensing platforms, providing an opportunity for citizen science. However, publicly available comprehensive sets of data are limited, and full workflow interfaces are missing. A prototype desktop application was developed to download, view, and process hyperspectral data. It allows browsing hyperspectral image libraries, examining spectra, assigning pixels to material classes, performing PCA, and automatic spectra classification. The USGS Search module provides an easy search engine to query the USGS repository of Hyperion hyperspectral images and seamlessly download them for analysis in the application. The new version of the application can perform more complex operations using Spectral Python, including displaying a class map for the image with an overlay showing each class with a unique color. The development of better workflows and interfaces for hyperspectral data can make this technology more accessible and usable for a wider range of scientists, practitioners, and even the general public.
Development of a Processing Workflow for Hyperspectral Images
Hyperspectral Images (HSI) are collection of grayscale images of the same scene taken over narrow light wavelength intervals. HSI have been used extensively in various fields such as earth sciences, manufacturing, agriculture, food safety, defense, and homeland security. Hyperspectral data has become more accessible due to the increasing number of sensing platforms, providing an opportunity for citizen science. However, publicly available comprehensive sets of data are limited, and full workflow interfaces are missing. A prototype desktop application was developed to download, view, and process hyperspectral data. It allows browsing hyperspectral image libraries, examining spectra, assigning pixels to material classes, performing PCA, and automatic spectra classification. The USGS Search module provides an easy search engine to query the USGS repository of Hyperion hyperspectral images and seamlessly download them for analysis in the application. The new version of the application can perform more complex operations using Spectral Python, including displaying a class map for the image with an overlay showing each class with a unique color. The development of better workflows and interfaces for hyperspectral data can make this technology more accessible and usable for a wider range of scientists, practitioners, and even the general public.