Kuniaki Uehara

A Motion Recognition Method by Using Primitive Motions

Contents-based retrieval of multimedia information has been investigated in several research projects. In this paper, we will focus on an automatic indexing method for human motion data. We convert a motion data, which is represented as time series of 3-D position, into a symbol sequence. We call this method as conversion automatic indexing. The automatic indexing is performed in a pattern matching approach. Reference patterns are necessary for pattern matching, so that we will propose two methods to define primitive motions in order to make reference patterns. The first method divides motion data into segmental motion data by detecting the change of motion speed. The second method classifies segmental motions such that similar segmental motions are gathered in the same cluster. In order to evaluate the similarity between two segmental motions, we use the Dynamic Time Warping (DTW) method because each segmental motion takes different time length even if the same person performed the same motions. Motion data can be converted into a symbol sequence which represents a sequence of primitive motions. Then, Continuous Dynamic Programming (CDP) method is used to recognize contents of motion. CDP is one of the extensions of DTW. It makes us possible to recognize a motion with ease even if it is complex.

A location-aware graphical BBS for mobile environments

We propose a graphical BBS which can handle the users current location. In the system, a BBS room is related to a real object. Each BBS comment has a valid time and an influence area decided automatically. The users location then influences the system based on the users movements to display comments on the users screen. The influence area changes according to external events such as emergencies to express the importance and area of effect of the comment at the present time. Furthermore, we discuss the implement issues of the prototype system.

Semantic Structures for Video Data Indexing

Video indexing based on contents annotations can fully explore semantic information of video data. However, the most difficult and time-consuming process in annotation-based indexing is to identify appropriate video intervals for various semantic contents manually. Thus, automatic discovering video intervals from video data will be helpful for the indexing work. For this purpose, we propose semantic structures of video data and a mechanism for discovering semantic structures. The basic concept of our approach is to (1) discover consecutive sequences of shots from video data, each of which represents a consistent action or situation, and (2) index each of the discovered video intervals based on its semantics. A semantic structure is a collection of discovered video intervals that are classified into three categories: unchanged (i.e. actors or backgrounds are unchanged throughout the interval), gradually changing (i.e. actors or backgrounds are changing shot by shot) and multiplexing (i.e. individual actors or backgrounds are appearing by turns). The mechanism discovers these types of video intervals by comparing and contrasting similarity between each shot, and indexes each of discovered intervals by using indexing algorithms prepared for each type. We show how well our approach works for identifying video intervals with some experimental results.

Creating multimedia presentation based on constraint satisfaction problems in multimedia databases

We have studied a mechanism for creating multimedia presentation from various multimedia objects such as video, image and text. The important points for creating multimedia presentation are specification and verification. Specification means representation for temporal relations among multimedia objects. Temporal relations specify the flow of multimedia presentation in space and time. As for verification, we should consider both syntactic and semantic aspects in creating multimedia presentation. Syntactic aspect means how multimedia objects are scheduled. A graph model visually indicates temporal relations between multimedia objects, and can simulate the multimedia presentation in advance. This simulation can detect contradictions between temporal relations and configuration errors. Semantic aspects correspond to the presentation constraints. Presentation constraints can be detected when user-specific presentation is played out because users may have various requirements. We propose the way to create multimedia presentation so as to satisfy the constraints between multimedia objects.

A Time-Stamped Authoring Graph for Video Databases

This paper presents a “time-stamped authoring graph”, which is a new video description model and retrieval method based on fragmentary descriptions of video contents (i.e., the authors impressions or feelings) and their relationships. It. can handle continuous video images better than existing description models, which tend to have problems in extracting parts of continuous video images and giving them precise descriptions. A video retrieval algorithm using a time-stamped authoring graph is explained, and the criteria for selecting or ranking the retrieved video images are discussed. The design and implementation of the prototype for video description and retrieval interfaces are also described and examined, with some examples.

Video Summarization Based on Semantic Representation

Summarization of video data is of growing practical importance because the more expanding video databases, inundate users with vast amounts of video data, the more users need reduced versions which they can assimilate with limited effort in shorter browsing time. In recent days, many researchers have investigated summarizing techniques, such as fast-forward play back, and skipping video frames at fixed intervals of time. However, all these techniques are based on syntactic aspects of the video. n nAnother idea is to present summarized videos according its semantic representation. The critical aspect of compacting a video is context understanding, which is the key to choosing significant scenes that should be included in the summarized video. The goal of this work is to show the utility of semantic representation method for video summarization. We propose a method to extract significant scenes and create a summarized video without losing the content of the videos story. The story is analyzed by its semantic content and is represented in a structured graph where each scene is represented by affect units.

Video description model based on prototype-instance model

Current video database systems are able to manipulate syntactic aspects of temporal media like video and audio. However, few of these systems are able to manipulate video information according to its semantic content. Furthermore, manually describing video content is tedious and not practical for the video data due to the temporal, spatial, and massive characteristic of the video, meanwhile content description always depends on user interpretation. Our video description model is based on object-oriented model based on prototypes which is proposed in the object-oriented programming field. A prototype is a prefabricated object with a predefined structure, content and behavior from which new objects can be constructed from other prototypes. By using prototypes, the user is able to describe video stories easily implementing and modifying prototypes of typical scenes that usually appear in several stories. Our method uses two hierarchical structures, one based on prototypes inheritance structure and another based on prototype internal structure. From this point of view, our system includes a hierarchical prototype menu, which guides the user to reach easily requested prototypes; hence, the story remains the same regardless of different user interpretation.

Texture analysis and classification using bottom-up tree-structured wavelet transform

One difficulty in texture analysis was the lack of adequate tools for characterizing textures. Recent developments in multiresolution analysis, such as the wavelet transform, attempt to overcome this difficulty. The tree-structured wavelet transform (TSWT), which affords for an analysis in terms of the texture’s decomposition structure and the corresponding spectrum in the space-frequency domain, has received a lot of attention. Recently, Chang [1] proposed a texture classification algorithm based on a top-down TSWT (TSWTTD). Although high classification rates were reported, its performance is highly dependent on a set of parameters which have to be determined through elaborate tuning. We propose a texture classification algorithm based on a bottom-up TSWT (TSWTBU), which is not dependent on ad-hoc parameters and shows also superior performance. We also provide an analysis of the structural stability of TSWT.

Knowledge Integration of Rule Mining and Schema Discovering

Despite the growing popularity of semi-structured data such asWeb documents and bibliography data, most data mining researches have focused on databases containing well structured data like RDB or OODB. In this paper, we try to find useful association rules from semi-structured data. However, some aspects of semi-structured data are not appropriate for data mining tasks.

Parallel network organization algorithm for graph matching and subgraph isomorphism detection

Data representations using graphs are very flexible and are used in a wide variety of fields. The development of algorithms to perform basic processing at high speeds is vital for detecting subgraphs with important meaning from a graph set and for searching for subgraphs which match a given graph. However, as the number of graphs in question increases, the computation costs required for processing rise dramatically. In this paper, the authors describe an algorithm which detects isomorphisms which have several graphs in common from among a set of labeled directed graphs and then organizes them into a network based on the detected isomorphisms. This algorithm also provides greater processing speed through heuristics using the MDL (Minimum Description Length) principle.