多视觉特征与K-centroid聚类的高分辨率遥感图像检索

High Resolution Remote Sensing Image Retrieval Based on Multi-Visual Feature and K-centroid Clustering

目前基于单一内容的高分辨率遥感图像检索具有描述片面、信息不精确的问题。针对此问题,充分利用遥感图像的颜色、形状和纹理特征,将三者综合起来,形成多视觉特征的遥感图像检索,并通过一系列的迭代运算,得到这三种特征对待不同类遥感图像时各占的最佳比例系数,从而得到较好的检索结果。并针对分别计算遥感图像的颜色、形状和纹理特征,再将其融合导致在大图像库中进行检索时检索速度较慢这个问题,引入改进的K-centroid聚类算法,先对遥感图像库中的图像进行聚类,大大缩小了检索的范围,提高了检索速度。实验结果表明,该方法具有较好的检索结果。

At present,high resolution remote sensing image retrieval based on single content has the problem of one-sided description and imprecise information. The color,shape and texture features of remote sensing images were fully used and combined to form multi-vision remote sensing image retrieval in order to solve this problem.Through a series of iterative operations,the best proportionality coefficient for these three features to treat different types of remote sensing images can be obtained,which gets a better search result. Aiming at the problem of the retrieval speed is slow when searching the large image database for the color,shape and texture features of the remote sensing image respectively,the improved K-centroid clustering algorithm which firstly clustered the images in the remote sensing image database is introduced to reduce the retrieval scope as well as improve the retrieval speed.The experimental results show that the method has good retrieval results.