论文部分内容阅读
根据JPEG2000对感兴趣区域优先编码,以及感兴趣区域的形状可随意选取的特点,提出了一种在JPEG2000压缩码流不完全解码的情况下,实现多谱段遥感图像感兴趣目标的检索方法,该方法利用了遥感图像的性质,根据例子图像的谱特征对感兴趣区域的内容进行分析,并设计了一套相似性度量的方法。实验结果表明,此方法有较理想的图像检索效果和很高的检索效率,解决了应用上对实时性的高要求与遥感图像库数据海量性之间的矛盾。
According to the characteristics of JPEG2000, which is the priority coding of the region of interest and the shape of the region of interest, which can be chosen at will, a retrieval method is proposed to realize the target of multi-spectral remote sensing image when the JPEG2000 compressed stream is not completely decoded. The method takes advantage of the nature of the remote sensing image, analyzes the content of the region of interest according to the spectral features of the example image, and designs a set of similarity measures. Experimental results show that this method has better performance of image retrieval and higher retrieval efficiency, and solves the contradiction between the high real-time requirement and the huge amount of data of remote sensing image database.