Publication Date:
2021-01-28
Description:
The oil spill creating potentially serious environmental impacts on both marine life and the coastal shorelines. Accurately oil spill monitoring can reduce economic loss and assess these impacts. With the development of imaging technology, high spectral resolution data in hyperspectral imagery (HSI) sensors provide a valuable source of information that can be used for oil spill area segmentation by semi-automatic methods. At present, there are many methods for oil spill segmentation, most of which are based on threshold or neural network. These methods can achieve better segmentation results when the oil spill image is clear, but do not effectively segment the oil spill area when the image with high noisy and the oil spill area is blurred. In this article, for hyperspectral images blurred with high noisy, a BF-MD-LBF model is proposed. There are two key steps in the proposed method: (1) To take advantage of spectral information, KPCA is introduced to Local Binary Fitting (LBF) energy function and a new energy function model is constructed; (2) To have hyperspectral image smoothed without blurring the edges, the bilateral filter is incorporated into the LBF energy function as regularization term.
Language:
English
Type:
info:eu-repo/semantics/conferenceObject
