Investigation of epifauna coverage on seagrass blades using spatial and spectral analysis of hyperspectral images
Authors:
- 1 June 2013
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-4
Abstract
Seagrasses are critical components of marine food webs, yet negatively impacted by excessive coverage of epiphyte (attached algae or plants) or epifauna (organisms or animals). Since excessive nutrients result in excessive epiphyte/epifauna growth, epiphyte/epifauna coverage on seagrasses may serve as an indicator of nutrient status of local marine systems. In this work, the hyperspectral images obtained from seagrass samples are investigated to quantify percent epifauna coverage. The N-FINDR algorithm is utilized for spectral analysis to identify endmembers. Each test spectrum is then correlated with endmembers for classification. The classification results (epifauna or not epifauna) are mapped on a 2D binary image. In spatial analysis, a single image frame from a series of hyperspectral image frames is selected and analyzed using a 2D image processing chain. We show that independent spectral and spatial analysis results produce comparable results.Keywords
This publication has 7 references indexed in Scilit:
- Fast GPU algorithms for endmember extraction from hyperspectral images2012
- Hyperspectral Unmixing Overview: Geometrical, Statistical, and Sparse Regression-Based ApproachesIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2012
- Spatial-Spectral Preprocessing Prior to Endmember Identification and Unmixing of Remotely Sensed Hyperspectral DataIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2012
- Linear Versus Nonlinear PCA for the Classification of Hyperspectral Data Based on the Extended Morphological ProfilesIEEE Geoscience and Remote Sensing Letters, 2011
- Detection of seagrass in optical shallow water with Quickbird in the Xincun Bay, Hainan province, ChinaIET Image Processing, 2011
- Hyperspectral Image Classification Using Dictionary-Based Sparse RepresentationIEEE Transactions on Geoscience and Remote Sensing, 2011
- N-FINDR: an algorithm for fast autonomous spectral end-member determination in hyperspectral data1999