基于区域增长迭代算法的SAR海冰图像分割

针对SAR图像分割仅考虑像素本身灰度值而不考虑空间特征的阈值法和聚类方法中存在图像有噪或边界模糊问题，提出一种基于区域增长迭代算法的SAll．海冰图像分水岭分割。在整个处理过程中保持对分割的细化并同时以迭代方式产生语义类标记。迭代算法的初始阶段应用K均值聚类法在二维特征空间中进行分割；迭代阶段包括Gibbs采样、自相关函数最小化和区域合并。自相关函数锐度指数最小化使图像更清晰。真实SAR图像的实验结果表明该算法的快速性和有效性，同时保留目标的细节信息。

SAR SEA ICE IMAGE SEGMENTATION BASED ON REGION GROWING AND ITERATED ALGORITHM

Aim at SAR image segmentation, we only consider the gray value of pixel itself without considering the spatial characteristics for threshold method and image noise or fuzzy boundary problems for clustering method. We propose a kind of SAR watershed segmentation algorithm for sea ice image based on region growing. We keep refining the segmentation and producing semantic class labels at the same time in an iterative manner in the overall process. Segmentation in the two-dimensional feature space by using K means clustering method for initial stage of iteration algorithm. Iterative phase includes Gibbs sampling, autocorrelation function minimization and region merging. Autocorrelative function sharpness index minimization makes the image more clearly. Experimental results on real SAR images show that the speed and effectiveness of the algorithm, preserving the details of target also.