基于视觉词袋方法的不同分辨率的遥感影像检索

Indexing of Remote Sensing Images with Different Resolutions Based on Bag of Visual Words Method

缩小语义鸿沟,是基于内容的遥感影像检索的必经之路,本文提出了一种基于多种底层特征的视觉词袋方法来进行不同分辨率的遥感影像的分类和检索。主要涉视觉词袋模型的构建以及该模型采用不同的底层特征描述对实验结果的影响。通过对不同分辨率的遥感影像进行基于不同底层特征组合的视觉词袋特征的提取,从而得到一系列分类检索实验结果。结果表明,基于底层特征组合的视觉词袋方法能有效地提高不同分辨率遥感影像的分类和检索效果。

How to narrow semantic gap of remote sensing images is only way to researchthe different resolution remote sensing image retrieval. In this paper,we present a method named Bag of Visual Words method. We use a variety of combinations of low-level features in Bag of Visual Words method. We mainly need to build a model of bag of visual words and to choose the best low-level features. In this paper we contrastthe classification result of different feature combination. The result show that the Bag of Visual Words method based on low-level features can effectively improve the correct rate of different resolution remote sensing image classification and retrieval.