QUERY BY IMAGE AND VIDEO CONTENT - THE QBIC SYSTEM
QUERY BY IMAGE AND VIDEO CONTENT - THE QBIC SYSTEM
复制标题
DOI：
10.1109/2.410146
复制
发表时间：
1995-09-01
期刊：
影响因子：
2.2
通讯作者：
YANKER, P
中科院分区：
文献类型：
--
作者：
FLICKNER, M;SAWHNEY, H;YANKER, P
扫描、网络、压缩和视频技术的进步以及多媒体计算机的普及导致了大量在线图像和视频的产生。这些集合产生了对定位特定图像或视频剪辑的新方法的需求。按图像内容查询（QBIC）项目正在研究通过按内容查询和检索图像和视频来扩展和补充基于文本的检索的方法。可以使用颜色、纹理、形状和对象位置等属性来执行检索。视频特定的查询包括那些对摄像机的运动参数，如变焦，摇摄，和物体motion.The项目已经导致了一个原型系统，两个主要步骤：数据库人口和查询。在人口中，方法识别静态图像中的对象，将视频分割成称为镜头的短序列，并计算描述颜色，纹理，形状，位置或运动信息的特征。在数据库查询中，图像和镜头可以通过示例（“显示与此图像相似的图像”）或通过从拾取器中选择属性（如色轮、草图形状、相机运动列表或这些的组合）来检索。关键的QBIC技术问题包括可视化查询语言和图形用户界面，该界面允许用户通过绘制、草图或选择图形元素来形成查询。关键还包括描述图像和视频内容的高维特征的索引技术、图像（识别有趣的对象）和视频（识别镜头和有趣的移动和静态对象）的自动分割技术以及相似性检索（匹配人类感知）。QBIC技术已进入Ultimedia和数字图书馆商业IBM产品。
Advances in scanning, networking, compression and video technology-and the proliferation of multimedia computers-have led to the generation of large on-line collections of images and videos. These collections have created a need for new methods to locate specific images or video clips. The Query by Image Content (QBIC) project is studying methods to extend and complement text-based retrievals by querying and retrieving images and videos by content.Queries can be performed using attributes such as colors, textures, shapes, and object position. Video-specific queries include those on camera motion parameters like zoom, pan, and object motion.The project has resulted in a prototype system with two major steps: database population and query. In population, methods identify objects in still images, segment videos into short sequences called shots, and compute features describing color, texture, shape, position, or motion information. In database query, images and shots can be retrieved by example (''Show me images similar to this image'') or by selecting properties from pickers such as a color wheel, a sketched shape, a list of camera motions, or a combination of these.Key QBIC technical issues include a visual query language and a graphical user interface that lets users form a query by painting, sketching, or selecting graphical elements. Key also are indexing techniques for high-dimensional features describing image and video content, automatic segmentation techniques for images (to identify interesting objects), and videos (to identify shots and interesting moving and static objects), and similarity retrieval (to match human perception).QBIC technology has moved into the Ultimedia and Digital Library commercial IBM products.