Mean-shift Segmentation Parameters Estimator (MSPE): A new tool for Very High Spatial Resolution satellite images

摘要

In this research we propose a new tool, called Mean-shift Segmentation Parameters Estimator (MSPE), to automate the selection of segmentation parameter values. Our algorithm is based on Mean-shift segmentation. However, this algorithm needs as inputs: i) existing vector database and ii) spectral data to define automatically the segmentation parameter values. The proposed method has been applied on different landscape's types (Urban, and Rural). The results obtained show accurate results.