基于边缘角度二阶差分直方图的城市遥感图像检索方法

Remote sensing image retrieval based on histogram of edge angular second-order difference

针对目前存在的遥感图像检索方法中,图像区域分割困难,不易准确提取局部形状特征的问题,提出一种无需进行区域分割的全局形状特征,即基于Freeman差分码思想的边缘角度二阶差分直方图特征;通过检索特征和传统的基于边缘角度直方图特征的遥感图像检索对比实验,揭示了该遥感图像检索方法具有更好的检索性能,检索精度提高了约32%。

To solve problems such as the difficulties of image region segmentation and local shape features extraction of the existing methods of remote sensing image retrieval , this paper proposed a global shape feature that requires no region segmentation, namely, the histogram feature of edge angular second-order difference based on Freeman difference code. Remote sensing image retrieval experiments were conducted to compare these retrieval features with the traditional histogram features based on edge angle. The results show that the retrieval performance of the proposed method is better than the traditional one and its retrieval precision increases by 32%.