Automated Synthetic Aperture Sonar Image Segmentation using Spatially Coherent Clustering

Seabed image segmentation is an important product for a variety of fields, including, habitat mapping, geological surveys, mine counter measures, and naval route planning. Developing a clustering algorithm that can both accurately segment high resolution imagery and generalize well over large areas is challenging. In this paper we will evaluate the performance of a new unsupervised image segmentation algorithm. The method utilizes imagery derived features (intensity and texture) to identify clusters (different seabed types) in feature space while also encouraging local homogeneity. In this paper we will demonstrate how our spatially coherent k-means clustering algorithm can efficiently and accurately segment Synthetic Aperture Sonar (SAS) images. Our experiments show that our spatially coherent clustering algorithm can increase segmentation accuracy up to 15 and 20 percent relative to OpenCV k-means and ArcGIS Pro ISO clustering, respectively.