摘要:
The present application discloses systems and methods for querying an image database. In one embodiment, a method is disclosed that includes maintaining in data storage a plurality of images and a plurality of geo-feature identifiers, where each image is associated with at least one geo-feature identifier that indicates at least one geographic cell in which the image was recorded and a visual feature of the image. The method further includes receiving a query image and an indication of a requested geographic location, determining at least one target geographic cell corresponding to the requested geographic location, and identifying at least one target visual feature within the query image. The method further includes determining at least one requested geo-feature identifier based on the at least one target geographic cell and the at least one target visual feature and selecting one or more images based on the at least one requested geo-feature identifier.
摘要:
Aspects of the invention pertain to matching a selected image/photograph against a database of reference images having location information. The image of interest may include some location information itself, such as latitude/longitude coordinates and orientation. However, the location information provided by a user's device may be inaccurate or incomplete. The image of interest is provided to a front end server, which selects one or more cells to match the image against. Each cell may have multiple images and an index. One or more cell match servers compare the image against specific cells based on information provided by the front end server. An index storage server maintains index data for the cells and provides them to the cell match servers. If a match is found, the front end server identifies the correct location and orientation of the received image, and may correct errors in an estimated location of the user device.
摘要:
Aspects of the invention pertain to matching a selected image/photograph against a database of reference images having location information. The image of interest may include some location information itself, such as latitude/longitude coordinates and orientation. However, the location information provided by a user's device may be inaccurate or incomplete. The image of interest is provided to a front end server, which selects one or more cells to match the image against. Each cell may have multiple images and an index. One or more cell match servers compare the image against specific cells based on information provided by the front end server. An index storage server maintains index data for the cells and provides them to the cell match servers. If a match is found, the front end server identifies the correct location and orientation of the received image, and may correct errors in an estimated location of the user device.
摘要:
Aspects of the invention pertain to matching a selected image/photograph against a database of reference images having location information. The image of interest may include some location information itself, such as latitude/longitude coordinates and orientation. However, the location information provided by a user's device may be inaccurate or incomplete. The image of interest is provided to a front end server, which selects one or more cells to match the image against. Each cell may have multiple images and an index. One or more cell match servers compare the image against specific cells based on information provided by the front end server. An index storage server maintains index data for the cells and provides them to the cell match servers. If a match is found, the front end server identifies the correct location and orientation of the received image, and may correct errors in an estimated location of the user device.
摘要:
An image retrieval technique employing a novel hierarchical feature/descriptor vector quantizer tool—‘vocabulary tree’, of sorts comprising hierarchically organized sets of feature vectors—that effectively partitions feature space in a hierarchical manner, creating a quantized space that is mapped to integer encoding. The computerized implementation of the new technique(s) employs subroutine components, such as: A trainer component of the tool generates a hierarchical quantizer, Q, for application/use in novel image-insertion and image-query stages. The hierarchical quantizer, Q, tool is generated by running k-means on the feature (a/k/a descriptor) space, recursively, on each of a plurality of nodes of a resulting quantization level to ‘split’ each node of each resulting quantization level. Preferably, training of the hierarchical quantizer, Q, is performed in an ‘offline’ fashion.
摘要翻译:采用包含分层组织的特征向量集的新颖的分层特征/描述符向量量化器工具“词汇树”的图像检索技术,其以分层方式有效地划分特征空间,创建映射到整数编码的量化空间 。 新技术的计算机化实现采用子程序组件,例如:该工具的训练器组件生成分层量化器Q,用于在新颖的图像插入和图像查询阶段中的应用/使用。 通过在特征(a / k / a描述符)空间上递归地在所得到的量化级别的多个节点的每个节点上运行k-means来“分割”每个结果的每个节点来生成分层量化器Q工具 量化级别。 优选地,以“脱机”的方式执行分层量化器Q的训练。