基于数据驱动的SAR溢油图像分割算法对比研究

本文针对SAR图像中溢油区域的散射特点,首次将二维最大类间方差阈值分割算法应用于SAR溢油分割,并提出了一种基于改进二维最大类间方差的SAR溢油图像分割算法。基于ENVISAT ASAR溢油图像的实验结果表明,和经典最大熵以及原始二维最大类间方差分割算法相比,本文算法是一种抗噪性能好,分割精度高,运算速度快的SAR溢油图像阈值分割算法。

On consideration of the scattering characters of the oil spil area in SAR image, this paper firstly applies the 2D-otsu threshold algorithm into the SAR image oil spil detection, and proposes an improved SAR image oil segmentation algorithm based on 2D-otsu. The comparisons results derived based on maximum entropy, 2D-Otsu and the proposed algorithm indicate that the improved one is a satisfied oil threshold segmentation method with better anti-noise performance, high segmentation accuracy and fast calculation speed.