A Novel Superpixel Generation Method Based on Edge Intensity for SAR Image
Yuan Chen, Lingjun Zhao, Tao Tang, Gangyao Kuang · 2018 Progress in Electromagnetics Research Symposium (PIERS-Toyama) · 2018
With the acquisition of high resolution SAR image data, more superpixel-based methods have emerged for SAR image analysis in recent years. The improvement of the existing SAR image superpixel generation algorithm has been a hot topic in academia. This paper presents a novel superpixel generation method based on edge intensity and linear spectral clustering for SAR image with speckle noise, which is named e-LSC. Experiments demonstrate that the proposed algorithm has some advantages in terms of accurate boundary adherence and regularization of homogeneous regions compared with other algorithms.