Journal article
A multi-scale segmentation approach to mapping seagrass habitats
pp.60-60
01 Mar 2005
Abstract
The purpose of this study was to map the areal extent and density of submerged aquatic vegetation within the Barnegat Bay and Little Egg Harbor, New Jersey as part of ongoing monitoring for the Barnegat Bay National Estuary Program. We examined the utility of multi-scale image segmentation/object-oriented image classification approaches to seagrass mapping. While the aerial digital camera imagery employed in this study had the advantage of flexible acquisition, suitable image scale, fast processing return times and comparatively low cost, it had inconsistent radiometric response across the individual images. While we were not successful in using the eCognition software to develop a rule-based classification that was universally applicable across the 14 individual image mosaics that comprised our 73,000 ha study area, the manual classification approach that we developed provided a flexible and time effective approach to mapping seagrass. From a theoretical standpoint, the multi-scale image segmentation/object oriented classification approach closely mirrored our conceptual model of the spatial structure of the seagrass beds and associated bottom features. Rather than visualizing the seagrass habitats as simply a collection of like pixels, this object oriented approach successfully extracted the spatial features of ecological interest. This multi-scale image segmentation approach coupled with field transect/point surveys has the potential to be more replicable than strictly boat-based surveys and/or visual image interpretation and allow for more robust conclusions regarding change in areal extent, location and spatial pattern of seagrass beds through time.
Metrics
1 Record Views
Details
- Title
- A multi-scale segmentation approach to mapping seagrass habitats
- Creators
- R LathropP MontesanoS Haag
- Publication Details
- pp.60-60
- Resource Type
- Journal article
- Language
- English
- Academic Unit
- Computer Science (Computing)
- Identifiers
- 991021862290804721