基于颜色和纹理特征的遥感图像检索

摘要

Remote sensing image retrieval based on texture features and color features is proposed,for meeting the limitations of single features of remote sensing image retrieval and the computational cost of traditional retrieval methods. After the analysis of the existing re-mote sensing image retrieval,the general frame of remote sensing image retrieval which is based on the color and Gabor wavelet texture features are established. Design a group of multi-scale and multi-directional filters,optimize and select the filter parameters. First of all, do image filtering,then conduct the texture feature extraction,finally extract color feature with color histogram for remote sensing image, eventually in fusion of these two kinds of image features,design and implement a image retrieval prototype system based on color feature and Gabor wavelet texture feature. The obtained color and texture features are used to retrieve image database. The experimental results show that the proposed method is efficient.%　　针对单一特征遥感图像检索的局限性和传统综合特征检索方法计算复杂度高的问题，提出了一种基于纹理特征和颜色特征的遥感图像的检索。在分析了现存遥感图像的检索方法后，得出了用颜色特征及Gabor小波提取出的纹理特征进行图像检索的总体思路。设计了一组多尺度和多方向的滤波器组，优化并选择滤波器组的各参数。首先对图像进行滤波，然后再进行纹理特征的提取，最后用颜色直方图对遥感图像进行颜色特征提取，最终将融合这两种图像特征，设计并实现一个基于颜色特征和Gabor小波纹理特征的图像检索原型系统。在进行图像检索时将纹理特征和颜色特征融合对图像库进行图像检索。由实验结果表明，所提出的方法是有效的。