Segmenting mangrove ecosystems drone images using SLIC superpixels
2019
Zimudzi, Edward | Sanders, Ian | Rollings, Nicholas | Omlin, Christian
Mangrove ecosystems play a very important ecological role on land–ocean interfaces in tropical regions. These ecosystems comprise of various tree species and aquatic animals, protecting the environment and providing a habitat that supports many living organisms including humans. The identification of image regions in mangrove ecosystems plays a significant role in ecosystem monitoring and conservation. Recent studies have suggested oversegmentation of colour images using superpixels as a solution to the segmentation of image regions. This study used the SLIC superpixel algorithm and k-means clustering to segment images taken from a camera mounted on a drone from a mangrove ecosystem in Fiji. The SLIC superpixel algorithm performed well to demarcate image regions with similar colour and texture information into patches and to use k-means for the segmentation of the whole image. These results lend support to the use of superpixel algorithms for the segmentation of mangrove ecosystems. Understanding how superpixels can be used for the segmentation of drone images will assist conservation efforts in mangrove ecosystems.
显示更多 [+] 显示较少 [-]