This document addresses descriptor technology for search and retrieval applications, i.e. for visual content matching in video. Visual content matching includes matching of views of large and small objects and scenes, with robustness to partial occlusions as well as changes in vantage point, camera parameters and lighting conditions. The objects of interest comprise planar or non-planar, rigid or partially rigid, textured or partially textured objects, but exclude the identification of people and faces. The databases can be large, for example broadcast archives or videos available on the internet. Such applications thus require video descriptors that enable matching with smaller descriptor sizes and shorter runtimes as compared to application enabled by single-frame (still image) descriptors (e.g. CVDS, ISO/ IEC 15938-13) in the video domain.
Compact descriptors for video analysis for search and retrieval applications:
— enable design of interoperable object instance search applications
— minimize the size of video descriptors
— ensure high matching performances of objects (in terms of accuracy and complexity)
— enable efficient implementation of those functionalities on professional or embedded systems.
This document provides a complementary tool to the suite of existing standards, such as ISO/ IEC 15938-13.