基于可视范围的图像检索方法

发布时间：2018-06-01 17:34

本文选题：可定位图像 + EXIF格式　；参考：《地球信息科学学报》2016年08期

【摘要】：图像是视觉所及环境的自然再现,具有表达直观、内容丰富的特性,其已成为GIS空间数据采集与应用中一种重要的可视数据源。随着硬件终端的高速发展,集成的传感器愈加丰富,多源传感器信息融合已成为移动GIS数据懫集的发展趋势。针对移动终端获取方位角偏差问题,本文重点阐述了基于移动终端的重力传感器与磁力传感器获取镜头拍摄朝向的算法,并基于EXIF图像格式实现了图像元数据、定位信息、方位信息一体化采集生成可定位图像。针对现有可定位图像以点代面建模方式的不足,本文以镜头视野范围AOV(Angle of View)模型为基础,提出了以图像元数据、位置信息以及方位信息构建图像可视域的方法,并采用多级网格方式对图像可视域面状特征建立空间索引,实现基于可视域的图像检索。在此基础上,针对福建省野外遥感核查采集的8022幅可定位图像,采用点建模及可视域建模2种方式建立检索图像集。从检索速度、检索结果2个方面进行对比分析,表明基于可视域图像检索方法不仅能检索出拍摄位置在检索区域内的图像,还可检索出拍摄位置在检索区域外但实际拍摄场景在检索范围内的图像。同时,通过多级空间网格索引可提升6.22-8.64倍检索速度。
[Abstract]:Image is the natural reproduction of visual environment, which has the characteristics of intuitive expression and rich content. It has become an important visual data source in the acquisition and application of GIS spatial data. With the rapid development of hardware terminals, integrated sensors become more and more abundant, and multi-source sensor information fusion has become the development trend of mobile GIS data collection. Aiming at the problem of azimuth deviation of mobile terminal, this paper focuses on the algorithm of obtaining shot orientation by gravity sensor and magnetic sensor based on mobile terminal, and realizes image metadata and location information based on EXIF image format. Orientation information integrated acquisition to generate locatable image. Aiming at the deficiency of the existing method of point-to-face modeling of localizable image, this paper presents a method of constructing the visual domain of the image based on the AOV(Angle of view model, which is based on the image metadata, position information and azimuth information. The multilevel grid method is used to build spatial index for the viewable features of the image, and the image retrieval based on the visual domain is realized. On this basis, 8022 locatable images collected by field remote sensing verification in Fujian Province are built by point modeling and visual modeling. Comparing and analyzing from two aspects of retrieval speed and retrieval result, it is shown that the method based on viewable image retrieval can not only retrieve the image of shooting position in the retrieval area, It can also retrieve images where the shooting location is outside the retrieval area but the actual scene is in the retrieval range. At the same time, the retrieval speed can be increased by 6.22-8.64 times by multilevel spatial grid index.
【作者单位】：福建师范大学地理科学学院;福建省陆地灾害监测评估工程技术研究中心;中国科学院地理科学与资源研究所资源与环境信息系统国家重点实验室;
【基金】：国家自然科学基金项目(41201417) 资源与环境信息系统国家重点实验室青年人才培养基金项目(O8R8B6E0YA) 福建省自然科学基金项目(2016J01186) 福建省测绘地理信息局局校合作科技项目(2016JX02)
【分类号】：P208

【相似文献】