归一化转动惯量特征的遥感影像检索

Remote sensing image retrieval based on normalized moment inertia characteristic

针对传统遥感影像检索方法大多存在精度不高、效率低下等不足的问题,该文提出了一种基于归一化转动惯性特征的遥感影像检索算法。该算法对经过脉冲耦合神经网络处理的二值影像序列进行计算,提取影像序列的归一化转动惯量特征;同时,利用马氏距离结合Pearson积矩相关法来度量各特征矢量之间的相似性,提高检索结果的正确率。实验结果证明,该算法有效地兼顾了影像的内容结构信息,不仅可以快速地进行检索计算,还能提高检索精度。

Aiming at the low accuracy and poor efficiency of traditional remote sensing image retrieval methods, this paper proposed a remote sensing image retrieval method based on normalized moment inertia characteristics. The binary image sequence, which was obtained by pulse-coupled neural networks, was calculated to extract normalized moment inertia characteristics; meanwhile, the Mahalanobis distance and the Pearson product-moment correlation method were combined to measure the similarity between the feature vectors to improve the accuracy of retrieval results. Experimental results showed that this method effectively considered the image structural informa- tion, which could quickly and efficiently provide accurate retrieval results.