1.1 Interoperable Image Search and Retrieval
This Technical Report specifies two things. The first is a framework for interoperability for still image search
and retrieval. The second identifies an architecture and the components in this framework, the linkages
between components, and which of these components and links are to be standardized in JPSearch.
The image search and retrieval framework will be determined by real use cases (tasks) and will leverage on
lessons learnt in the long history of text retrieval where, for example, different users issuing the same query
may be looking for (very) different results. This is important because it means that the framework must be
general enough to support many possible approaches to image retrieval, e.g., from using only low-level image
features, to text annotations, to community input, or a mixture of such approaches.
From the framework and components, and the linkages and flow of data between them, the parts of JPSearch
that need to be standardized can be determined.
1.2 Motivation
There are many applications that provide image search and retrieval functionality on computer desktops, on
the World Wide Web (i.e., websearch), on imaging devices, and in other consumer and professional
applications. These implementations are characterized by significant limitations, including:
? Lack of the ability to reuse metadata
? Lack of a common query format and search semantics
? Lack of a common format for handling context in searching
1.3 Outline of the Technical Report
There will be 7 clauses to this report. They are arranged as follows:
Clause 2 provides definitions of terms and abbreviations.
Clause 3 reviews the traditional approaches to image search and from various examples, motivates the
importance of the user in the search process, the importance of making explicit the user task and user
evaluation, and that meaning in images may be as much added from outside as extracted from the image
itself. This motivates the next clause, which are the real use cases.
Clause 4 describes the real use cases of the various ways searching does take place. In particular, there
could be multiple entry points into the overall search framework. These would include automatic, semiautomatic,
and human (user) driven searches. These would include specific use cases and motivating
examples of searches for images.
Clause 5 describes the overall search and management process. This can be considered the requirements
specifications for a general search and management architecture.
Clause 6 describes the 4-layer architecture for JPSearch and explicitly identifies the components in the
architecture, and what their roles and positions are in the architecture. We will describe how the use cases
described in Section 3 map to the layers of this architecture.
ISO/IEC TR 24800-1:2007 history
2012ISO/IEC TR 24800-1:2012 Information technology - JPSearch - Part 1: System framework and components
2007ISO/IEC TR 24800-1:2007 Information technology - JPSearch - Part 1: System framework and components