针对传统局部信息模糊C均值聚类算法权重系数仅由像素间欧式距离决定,无法准确衡量和充分利用像素间的相似性,对SAR图像分割不准确的问题,提出了一种全新的局部信息相似性描述方法,并结合图像的非局部信息,对像素到聚类中心的距离和像素隶属度计算方法进行改进,并提出了一种同时包含图像局部和非局部信息的改进SAR图像分割方法。实验表明,与其他模糊聚类方法相比,该方法在抑制SAR图像相干斑噪声的同时,能较好地保护SAR图像目标的边缘和细节,具有很好的SAR图像分割效果。

Aiming at the problem that the weight coefficients of traditional local information fuzzy C-means clustering algorithm are only determined by the Euclidean distance between pixels, which cannot accurately measure and fully utilize the similarity between pixels and is inaccurate for SAR image segmentation, a new similarity description method of local information is proposed, and the distance from pixels to the cluster center and the pixel affiliation calculation method are improved by combining the non-local information of images, and an improved SAR image segmentation method that contains both local and non-local information of the image is proposed. The experiments show that compared with other fuzzy clustering methods, the method in this paper can better protect the edges and details of SAR image targets while suppressing the coherent speckle noise of SAR images, and has a good SAR image segmentation effect.

TN957. 52

国家自然科学基金资助项目(No. 61771411);特珠环境机器人四川省重点实验室项目(13zxtk08);博士基金资助项目(17zx7159)