论文部分内容阅读
针对基于形状的遥感图像检索中区域分割困难这个问题,本文提出利用边缘角度直方图作为形状特征对遥感图像进行检索;针对遥感影像分辨率和颜色反差对检索性能的影响,本文提出采用多尺度分析的思想,提取图像在多个尺度上的边缘角度直方图特征进行检索。通过单一尺度边缘角度直方图特征和多尺度边缘角度直方图特征检索的对比实验,揭示了本文所提出的改进方案使遥感图像的检索精度提高了约31个百分点。
In view of the difficulty of region segmentation in shape-based remote sensing image retrieval, this paper proposes the use of edge-angle histogram as shape feature to retrieve remote sensing images. In view of the influence of remote sensing image resolution and color contrast on retrieval performance, this paper proposes a multi-scale analysis The idea of extracting the histogram features of the edges at multiple scales is retrieved. The comparative experiments of single-scale edge-to-edge histogram feature and multi-scale edge-to-edge histogram feature retrieval reveal that the improved scheme proposed in this paper improves the retrieval precision of remote sensing image by about 31%.