论文部分内容阅读
随着遥感图像数量的急剧增加,如何进行高效检索已成为遥感图像信息提取和共享的瓶颈问题,基于内容的遥感图像检索因此逐渐成为了研究热点。本文提出了一种新的遥感图像检索方法,该方法综合利用了图像的色调和纹理特征。其基本过程是:首先,对图像进行主成分变换,对变化后的第一主成分图像进行五叉树分解,将大幅面的遥感图像分成一系列的子图像;然后,利用多通道Gabor滤波器与子图像做卷积运算,提取其纹理特征,同时计算像元值的方差和三阶矩作为各子图像的色调特征;最后,以子图像为特征基元,构建图像的色调直方图和纹理直方图,以多特征直方图匹配方法计算图像相似度实现遥感图像检索。利用高分辨率遥感影像的检索实验证明该方法是有效的。
With the dramatic increase in the number of remote sensing images, how to efficiently retrieve them has become a bottleneck in the extraction and sharing of remote sensing image information. Therefore, content-based remote sensing image retrieval has gradually become a research hotspot. In this paper, a new remote sensing image retrieval method is proposed, which takes full advantage of the hue and texture features of the image. The basic process is as follows: Firstly, transform the principal component of the image, decompose the transformed first principal component image by a pent tree, and divide the large-scale remote sensing image into a series of sub-images; then, use a multi-channel Gabor filter And the sub-image convolution operation to extract the texture features, while calculating the pixel value variance and third-order moments as the sub-image color tone features; Finally, the sub-image features as the feature base to build the image of the hue histogram and texture Histogram, Multi-feature histogram matching method to calculate image similarity for remote sensing image retrieval. The retrieval experiment of high resolution remote sensing image shows that this method is effective.