Jacques Kouloumdjian

A database approach for modeling and querying video data

Indexing video data is essential for providing content-based access. In this paper, we consider how database technology can offer an integrated framework for modeling and querying video data. As many concerns in video (e.g., modeling and querying) are also found in databases, databases provide an interesting angle to attack many of the problems. From a video applications perspective, database systems provide a nice basis for future video systems. More generally, database research will provide solutions to many video issues, even if these are partial or fragmented. From a database perspective, video applications provide beautiful challenges. Next generation database systems will need to provide support for multimedia data (e.g., image, video, audio). These data types require new techniques for their management (i.e., storing, modeling, querying, etc.). Hence, new solutions are significant. This paper develops a data model and a rule-based query language for video content-based indexing and retrieval. The data model is designed around the object and constraint paradigms. A video sequence is split into a set of fragments. Each fragment can be analyzed to extract the information (symbolic descriptions) of interest that can be put into a database. This database can then be searched to find information of interest. Two types of information are considered: (1) the entities (objects) of interest in the domain of a video sequence, and (2) video frames which contain these entities.

An advanced knowledge base management system based on the integration of logic programming and relational databases

An interface implemented in the environment between PROLOG and a relational DBMS is described. In order to combine the advantages of interpreted and compiled coupling strategies, the hybrid interface uses a two-phase communication protocol. This strategy minimizes the accesses to the database and generates queries as selectively as possible to the database by maximal instantiation of the database predicates, using DBMS for relational operations whenever possible. This interface is integrated into EPSILON (a knowledge-base management system) by defining it as a new inference engine which allows the user to use logical predicates and database predicates in the same way.<<ETX>>

A genetic model for video content based retrieval

Recent progress in compression technology has made it possible for computer to store efficiently pictures, audio and even video. Nevertheless, if such media are widely used in todays communication, efficient computer exploitation is still lacking because automated, high-level interpretation is nowadays impossible. While storing raw video streams in analogic or digital formats is possible, to archive efficiently video documents in order to enable efficient retrieval implies being able to store and manage video content (e.g., raw features, human interpretation of the video document). Recently, this field has received a particular attention. In the past years, two approaches have been considered: (1) systems that use only automated feature extraction, like Jacob [4], and (2) more elaborated systems for concrete applications [3] [1] [5] that integrate user specifications in their video data model.

Integrating database technology and logic programming paradigm

The EPSILON logic programming environment allows the structuring of large knowledge bases expressed in logic languages using so called theories, allowing in parallel the integration of different logic languages. An interesting feature of the EPSILON framework is the ability to use data stored in a relational database from a logic program. One of the protocols used in the current prototype to access databases from logic theories is the kernel-DBMS interface, called hybrid because it combines the advantages and eliminates the drawbacks of interpretative and compiled coupling techniques.

Translating Relational Queries to Object-Oriented Queries According to ODMG-93

With the advent of object-oriented database systems there is a need to migrate/integrate old systems, mainly relational ones, to/into object-oriented systems. In this paper we propose a method to automatically translate SQL queries to OQL ones. The method is based on a graph representation of SQL queries. This graph is enriched and transformed in a series of stages which take into account the caracteristics of the object schema with respect to the relational one (new entity type, ISA relationship, class hierarchy, etc.). This semantic enrichment allows an optimized translation of the SQL queries compared with a purely syntaxic translation.

Modeling and querying video databases

Indexing video data is essential for providing content based access. This paper develops a data model and a rule-based query language for video content based indexing and retrieval. The data model is based on the notion of generalized strata, which can be seen as a set of intervals. Each interval can be analyzed to extract symbolic descriptions of interest that can be put into a database. This database can then be searched to find information of interest. Two types of information are considered: the entities (objects) in the domain of a video sequence; and video frames, called generalized strata, which contain these entities. To represent this information, our data model allows facts as well as objects and constraints. We present a declarative, rule-based, constraint query language that can be used to infer relationships about information represented in the model. In the video applications we are interested in, we wish to construct new generalized strata from old ones. To do this, our language has an interpreted function term (i.e., constructive term) to concatenate generalized strata. The language has a clear declarative and operational semantics.

Using artificial intelligence techniques for intelligent simulation in memory re-education

The increasing complexity of new applications has made it necessary to have tools offering a large spectrum of functionalities and several paradigms. This is especially true for intelligent tutoring systems which need (meta)rule based inference for adaptive teaching strategies, object oriented paradigm for learner modeling, graphic simulation tools and large data management facilities for storing lessons and results. The carrying out of such tools, where several software cooperate, should take into account the development of user-friendly environments. This paper describes an intelligent rehabilitation system used for people suffering from memory troubles. It is aimed at simulating everyday “ecological” situations by means of graphical scenes: patients are then asked to realize specific actions on the displayed objects such as preparing breakfast according to dynamically predefined scenarii. The construction of such a system needs to use intelligent tools, which are able to substitute neuropsychologists in the re-education phase. Our solution consists of combining Artificial Intelligence tools and Simulation ones. The system articulates around an expert system generator and an interactive graphical simulation package. The detailed architecture and the working of the system are described here.