DISTRIBUTED MULTIMEDIA DATA MANIPULATION

(1)讨论了基于CORBA的并行图像卷积处理的实现。使用CORBA使利用异构工作站集群成为可能，每个工作站都具有不同级别的计算能力。我们还提出了一个适合高效图像处理的工作站数量的分析模型，并报告了一些实验结果。(2)我们提出了一种重用由图像处理专家设计的图像处理程序实现操作的方法。我们称这个过程为“图像处理的历史”。因此，图像处理的历史可以看作是半结构化数据。我们将半结构化数据模型应用于图像处理的历史记录，并检查了管理和向历史记录显示查询的方法的有效性。(3)以实际医疗数据为例，研究了一种基于区间的时态数据表示和运算方法。本文还研究了时态数据库中的索引技术。数据分析和逻辑数据设计先于基于表示时态数据的两种替代方法的两个关系数据库的实现。本文还从存储效率和检索查询结果所花费的时间方面比较了这两种替代方法。

(1) We discussed implementation of parallel image convolution processing based on CORBA. Employing CORBA makes it possible to exploit a cluster of heterogeneous workstations, each of which has a different level of computing power. We also presented an analytical model for the number of workstations appropriate for efficient image processing, and reported some experimental results.(2) We have proposed a method for reusing the procedure for implementing operations for image-processing programs designed by image processing specialists. We call this procedure "histories of image processings". Thus, histories of image processings can be regarded as semi-structured data. We applied a semi-structured data model to histories of image processings, and we examined the effectiveness of a method for managing and presenting queries to the histories.(3) An interval-based approach to represent and to perform operations on temporal data was examined using a real medical data. Indexing techniques in temporal databases was also investigated. Data analysis and logical data design precede the implementation of two relational databases based on the two alternatives approaches of representing temporal data. A comparison of the two alternatives in terms of their storage efficiency and time taken to retrieve results of queries is also undertaken.