基于上下文敏感的贝叶斯网络及方向关系的遥感图像检索

发布时间：2018-06-25 04:20

本文选题：图像检索 + 上下文敏感的贝叶斯网络(CSBN)　；参考：《国土资源遥感》2017年03期

【摘要】：如何在遥感图像数据库中快速、准确地找出目标图像,是检索系统的核心所在。基于对上下文语境敏感的贝叶斯网络(content-sensitive Bayesian network,CSBN),建立了含有方向关系的检索模型,并根据城市区域的特点,提出了适合城市区域检索的方法。首先,通过贝叶斯网络对图像进行检索;然后,依据图像的平均高频信号强度(average high frequency signal strength,AHFSS)对候选图像进行筛选;最后,得到含有城市区域这一高级语义特征的最终检索结果。为了确定图像内部的方向关系,采用东北、西北、东南和西南4个区域的方向描述图像的8种方向关系,有效降低了算法的时间复杂度。实验结果表明,该方法可有效地描述图像的场景语义,并具有较高的查准率和检索效率,可满足用户的需求。
[Abstract]:How to find the target image quickly and accurately in the remote sensing image database is the core of the retrieval system. Based on context-sensitive content-sensitive network, a retrieval model with directional relationship is established. According to the characteristics of urban area, a method suitable for urban area retrieval is proposed. First, the image is retrieved by Bayesian network; then, the candidate image is filtered according to the average high frequency signal intensity of the image (average high frequency signal str. Finally, the final retrieval result with the advanced semantic feature of urban area is obtained. In order to determine the direction relation within the image, the four directions of northeast, northwest, southeast and southwest are used to describe the eight directions of the image, which effectively reduces the time complexity of the algorithm. The experimental results show that this method can effectively describe the scene semantics of the image, and has a high precision and retrieval efficiency, and can meet the needs of users.
【作者单位】：中煤地西安地图制印有限公司;兰州交通大学测绘与地理信息学院;甘肃省地理国情监测工程实验室;
【基金】：国家自然科学基金地区基金项目“面向空间方向关系查询的多模型复合方法研究”(编号:41561090) 甘肃省财政厅基本科研业务费项目“群组目标空间方向关系计算模型研究”(编号:214146) 甘肃省高等学校基本科研业务费项目“基于空间关系敏感的高分辨率卫星图像检索技术研究”(编号:213049) 中国博士后科学基金资助项目“基于高分辨率遥感影像的滑坡自动提取方法研究”(编号:2014M552558XB)共同资助
【分类号】：TP751

【相似文献】