Takayuki Kunieda

Image retrieval using spatial intensity features

The rapidly increasing usage of multimedia environments has led to a greater demand for image retrieval. Until recently, this retrieval has been done using keywords, which are generally added by hand to the image, and is thus inefficient. Recent studies concern the automatic extraction from the images of features that can then be used for retrieval. Humans often identify images by the colors and objects they contain, however these are insufficient for image retrieval. Humans also use structural information to identify images. Our studies focus on the structural information of the image, that is, the spatial texture feature and the spatial edge feature which consists of the primitive feature and its distribution. These features enable us to characterize images without color features and object extraction. In this paper, we show how to extract these features and how to compare the features extracted from two images. We carry out experiments on the retrieval of similar images and show the results.

Package-Segment Model for movie retrieval system and adaptable applications

We verify the Package-Segment Model which we are proposing as a representation method for multimedia content logical structure. We describe the effectiveness of the content logical structure. The Package-Segment Model consists of various objects that are defined as structure components and are involved in the construction processes. We developed an experimental retrieval system of the contents that are indexed by this Package-Segment Model. We confirm that the Package-Segment Model has representation flexibility and its object framework and retrieval mechanism are very adaptable. Therefore it can be integrated into various multimedia content management systems. A prototype application of automatic indexing and retrieval is also presented and evaluated.

XML-schema Dynamic Mapped Multimedia Content Description Tool

MovieTool is a Description tool for multimedia content with XML oriented representation. MovieTool can generate a content description file in the format defined MPEG-7, which is known as the multimedia content description interface. We adopted XML-Schema dynamic mapping mechanisms for the multimedia content description schemes. As a result, MovieTool can provide the flexibility and general versatility for various meta-data standards not only MPEG-7 but also B-XML etc.

Video Content Management Using Logical Content

A common video file format is generally used in video content management system composed of modules for archiving, retrieving, and video streaming on a network. However, the file format may not be suitable for these functions. This paper presents a new video content management that uses logical content. Logical content makes it possible to manage a system that may use files of differing formats to store the same content to be used by different modules. Using logical content makes it possible to access physical video files regardless of their formats, and to easily manage video content consisting of multiple physical files. Other modules are able to request the retrieval of a physical file by specifying a logical content ID and its range. Consequently, this system can be designed to be both modularized and extended. We also present an experimental video content management system that verifies the effectiveness of our video content management system.

Extended Package-Segment Model and Adaptable Applications

This paper aims to present and verify the Extended Package-Segment Model. It provides a more flexible representation of logical structure of multimedia content than the Package-Segment Model (PS-Model) which we have already proposed. The PS-Model structure represents recursively ways to divide content. It is suitable to represent logical structure of content generated by content-based analysis. However, there are also some restrictions in generating the PS-Model structure. To solve these problems, we have reconsidered the generation rules and here propose a new data model called the Extended Package-Segment Model (EPS-Model). The EPS-Model structure can represent scenes and cuts that are focused. Moreover the EPS-Model inherits the advantages of the PS-Model framework such as indexing, browsing, and retrieval mechanisms. We introduce two experimental systems based on the EPS-Model structure. We verified that this EPS-Model has flexibility and suitability of representation of the structure for multimedia content in any application.

Retrieval and indexing methodology for multimedia content descriptions

Description Methodologies for multimedia content are gaining in interest. MPEG-7, which has been authorized by the International Organization for Standardization (ISO) is specifically targeted for such content. We developed an effective index structure for multimedia content descriptions such as MPEG-7. We developed a useful indexing method, called the position locator index that searches for the location of the description part in the description. Still image and full text search engines and most suitable feature indexes can be used without modification. Since these indexes are optimized for each requirement, they provide high retrieval performance. We developed an experimental retrieval system that combined these feature indexes and the position locator index. We evaluated the performance of the retrieval by using a linear-tracing search and the position index for the content descriptions. The result was satisfactory and retrieval speed was 200 times faster than linear tracing the descriptions, and the index size could be kept at a minimum.