Hyperspectral data representation for signal enhancement
As the advance of the technology in sensor design and manufacture, hyperspectral imagers have become available since late of 80s. A typical hyperspectral data cube is composed of about 100 to 200 spectral measurements and provides rich spectral information of the ground cover materials, in general. However, not all the measurements are important and essential for individual applications where users have their focused culture or natural classes of interest. The use of the unnecessary measurements creates difficulties in machine learning processing and reduces the reliability of thematic mapping. Another problem is that the original measures may not be the distinct features. Data representation is required to generate better signatures of the classes of interest. In this project, supervised feature selection and generation will be investigated. The outcome is significant practically and scientifically.
Description of Work:
- Examine the materials spectral characteristics quantitatively
- Evaluation of current feature selection and generation methods
- Develop new effective techniques
- Implement and validation