Dr. Bin Liu
Project Description and Objectives
Superpixel generation is a widely accepted idea for synthetic aperture radar (SAR) image understanding, which uses the spatial context of images. The superpixel is the elementary unit in the following processes, and can provide additional information for image understanding. Currently, superpixel generation for SAR images is still an open problem. In this project, the student will develop a superpixel generation method for SAR images, which exploits the inherent statistical information of SAR images. The method should be suitable for both homogeneous and heterogeneous areas in the image. Particularly, the method should provide competitive performance in the heterogeneous areas, considering specific applications.
Understanding statistical modelling for SAR images
Developing a superpixel generation method for SAR images based on the statistical information
Evaluating the performance of the developed superpixel generation method, especially in the heterogeneous areas