Statistical Region Merging Segmentation
Remote sensing image segmentation is the critical process in the workflow of object-based image analysis. Recently, region merging methods have attracted growing attention because they are able to utilize more features than spectral signals derived from initial segments. Statistical Region Merging Segmentation However, the existing algorithms commonly use fixed parameters to control the process of region merging, which limits the possibility of the co-existence of large and small segments. To address this issue, we propose a self-adapted region merging method, based on spectral angle threshold, toward segmenting remote sensing images. This method involves two steps: i) multiband watershed transformation to initiate primitive segments and ii) self-adapted threshold-based region merging. The performance of the proposed algorithm is evaluated in a farmland division and compared to the existing region merging method implemented in SAGA. The results reveal the proposed segmentation method outperforms the SAGA method, as indicated by its lower discrepancy measure.