朗伯定律的宽观测带SAR海冰图像分割

Wide-swath SAR ice images segmentation based on Lambert's law

入射角效应是宽观测带SAR海冰图像分割的主要障碍之一。基于宽观测带SAR海冰图像数据,提出了一种集成余弦朗伯定律的分割算法。为了提高分割算法对SAR海冰图像的适应性,充分考虑了斑点噪声和入射角效应因素,并在区域K-means聚类、朗伯定律校正之后,进行区域合并。分别针对合成SAR海冰图像和星载SAR海冰图像的实验结果表明,该算法可有效提高分割的准确性。

Incidence angle effect of the SAR images is a major obstacle to the automatic interpretation of SAR sea ice image. Based on wide-swath SAR ice data, this paper proposes a new segmentation algorithm which integrates Lambert’s law correction step. The segmentation algorithm considers the effects of speckle noise and the angle of incidence of factors. The Lambert’s law correction and region merging will be combined. The efficiency of the proposed method has been demonstrated on the segmentation of synthetic SAR sea ice image and gulf of Bothnia SAR sea ice image, where the segmentation accuracy has been substantially improved in contrast to area-based Markov random field（MRF） algorithm.