Masayoshi Teraguchi

Learning personalized video highlights from detailed MPEG-7 metadata

We present a new framework for generating personalized video digests from detailed event metadata. In the new approach high level semantic features (e.g., number of offensive events) are extracted from an existing metadata signal using time windows (e.g., features within 16 sec. intervals). Personalized video digests are generated using a supervised learning algorithm which takes as input examples of important/unimportant events. Window-based features are extracted from the metadata and used to train the system and build a classifier that, given metadata for a new video, classifies segments into important and unimportant, according to a specific user, to generate personalized video digests. Our experimental results using soccer video suggest that extracting high level semantic information from existing metadata can be used effectively (80% precision and 85% recall using cross validation) in generating personalized video digests.

Personalized delivery of digest video managed on MPEG-7

Presents a framework to generate digest video clips from contents managed by meta-data based on MPEG-7, which are personalized by profiles of an individual user. Mobile users request to receive condensed information instantly, so that it is important to generate short video clips from a long image sequence. Unstructured video contents should be therefore managed by standard meta-data. First, we propose a method to assign values that are changed in a time interval of indexes that present the content of the scene. Next, content profiles are estimated so as to generate typical digest video clips to meet similar demands from many users. In conclusion, our approach does not generate completely personalized video for all users, but provides some typical digest video clips for users without well-learned user profiles to enable them to choose their favourite ones, which can be combined in individual requests from users.

Optimized web services security performance with differential parsing

The focus of this paper is to exploit a differential technique based on the similarities among the byte sequences of the processed SOAP messages in order to improve the performance of the XML processing in the Web Service Security (WS-Security) processing. The WS-Security standard is a comprehensive and complex specification, and requires extensive XML processing that is one of the biggest overheads in WS-Security processing. This paper represents a novel WS-Security processing architecture with differential parsing. The architecture divides the byte sequence of a SOAP message into the parts according to the XML syntax of the message and stores them in an automaton efficiently in order to skip unnecessary XML processing. The architecture also provides a complete WS-Security data model so that we can support practical and complex scenarios. A performance study shows that our new architecture can reduce memory usage and improve performance of the XML processing in the WS-Security processing when the asymmetric signature and encryption algorithms are used.

Rule-based XML Mediation for Data Validation and Privacy Anonymization

XML mediation for data validation and privacy anonymization of very large complicated XML messages defined in some industry-specific specifications such as HL7 is becoming increasingly important in SOA because a lot of applications depend on various kinds of hard-coded data validation and privacy anonymization functions, which makes it difficult to keep consistency of the functions among the applications in SOA when the schema of the XML messages in the specifications is updated. This paper proposes a uniform rule-based approach to realize the functions as an XML mediation separated from the applications, which makes it easy to maintain consistency of the functions even when they are updated in accord with the changes to the specifications. Therefore, application developers can readily utilize the mediation with many applications in SOA without additional modification to the applications. Our approach allows the developers to define a set of rules that consist of two components: 1) constraint conditions in a conceptual data notation in the XML message, and 2) actions performed only when the conditions are satisfied. In order to make the rules independent from both the implementation-specific data representation and the industry-specific knowledge, we automatically transform the rules into the implementation-specific data representation using two more factors; one is the data mappings from the data notation in the rules to the concrete data representation in the implementation, and the other is the implied data relationships hidden in the rules. It is very important to take into consideration a general way to import the implied knowledge because it often depends on the industry-specific data structure and it is usually given outside of the mediation system.

Rapid generation of event-based indexes for personalized video digests

This paper presents a novel video indexing method for providing timely personalized video delivery services. Several previous research reports have dealt with automatic content-based indexing. However, these systems require a lot of time to manually correct the indexes after preliminary automatic processing, making it difficult to use these indexes in practical services. In our method, reliable content-based indexes can be automatically generated by manually flagging predefined and easily recognizable events while watching the video (without rewinding). The generated indexes use temporal functions of significance based on some parameters predefined for each kind of flagged event. Therefore, we can reduce the time required for flagging and achieve rapid semi-manual video indexing. Finally, we confirm effectiveness of the indexes transformed into a subset of MPEG-7 descriptions in a practical service to automatically construct personalized video digests.

Frugal signal control using low resolution web-camera and traffic flow estimation

Due to rapid urbanization, large cities in developing countries have problems with heavy traffic congestion. International aid is being provided to construct modern traffic signal infrastructure. But often such an infrastructure does not work well due to the high operating and maintenance costs and the limited knowledge of the local engineers. In this paper, we propose a frugal signal control framework that uses image analysis to estimate traffic flows. It requires only low-cost Web cameras to support a signal control strategy based on the current traffic volume. We can estimate the traffic volumes of the roads near the traffic signals from a few observed points and then adjust the signal control. Through numerical experiments, we confirmed that the proposed framework can reduce an average travel time 20.6% compared to a fixed-time signal control even though the Web cameras are located at 500 m away from intersections.