Imaging Spectroscopy for Scene Analysis by Antonio Robles-Kelly & Cong Phuoc Huynh

Author:Antonio Robles-Kelly & Cong Phuoc Huynh
Language: eng
Format: epub
Publisher: Springer London, London

The uni-modal segmentation algorithm can recover the full extent of the absorption feature. Nonetheless, it may fail to perform well in cases where the continuum is not convex-shaped. Examples of this are the results on Eucalyptus leaf powder shown in Fig. 7.22. Furthermore, the uni-modal assumption on the absorption features may hold for mixed absorptions, as in the case of the Navel leaf spectra shown in Fig. 7.22. Here, the two absorption features, one shallow segment around 600 nm and the deep band around 670 nm, were mistakenly detected as a single absorption by the uni-modal segmentation algorithm. The reason is that the absorption around 670 nm is too strong and it makes the shallow band less prominent. As a result, their mixture forms only one peak after continuum removal and cannot be separated simply based on the uni-modal constraint. From our comparison results, we can assert that there is no single method which consistently give optimal results for all the spectra in the datasets. In our experience, the uni-modal segmentation algorithm performs better for the mineral spectra in the USGS spectral library. This is due to the fact that, for mineral spectral data, the cusps and valleys in the spectra are more obvious. For organic material, like plants and leaves, whose absorption features are “shallow”, the MMWT is, in general, a good option.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.