COFIN 2002

COFIN 02 De Natale DIT-PRJ-02-013

Status NOT active project
DISI role Partner
Project type Research Project
Dimension National
Acquisition date 2003-01-01
Start date 2003-01-01
End date 2004-12-31
SAP code 40100551

Project details

Project astract The project is targeted to the study and implementation of an efficient system for archiving and retrieval of multimedia information acquired by a multi-sensorial system.<br/>The main interest is to develop an effective codec module for audio-visual data, featuring an appropriate interface with the transmission channel (problem of video stream scalability). Furthermore, PER2 will develop some parts of the advanced query systems used to access the media server. The basis functions expected from the archiving system are summarized hereafter:<br/>- ability of receiving and storing multimedia streams coming in parallel from different sensors, possibly with different characteristics (e.g., static and mobile sensors, wireless/wired connection, analog/digital acquisition, …)<br/>- compressed-format data archiving with quality/channel constraints (e.g., possibility of accessing the data at various quality/detail levels).<br/>- access to stored data through via different terminals (e.g., workstations in the control room, mobile terminals with different graphical power).<br/>- simple and fast access to multimedia data (indexing & retrieval functions)<br/>- use of data in different applications with adequate quality (e.g., presentation to human operator, automatic face recognition, automatic tracking of persons and objects, etc.)<br/>In order to match these ambitious objectives, it will be necessary to exploit and integrate several state-of-the-art technologies, as well as to develop some innovative methodologies in the field of compression and archiving techniques. From the technological standpoint, the system core will be developed by carefully taking into account the recommendations of the standardization bodies in the field of multimedia signal representation and compression.<br/>A clear advantage in complying with such recommendations relies in the possibility of exploiting the compatibility and interoperability of the resulting system, as well as to exploit the availability of software (and possibly hardware) tools to achieve a more efficient process. Furthermore, the reference standards contain some very interesting features for the target scenarios of the project. Concerning the MPEG-4 standard, which will be the kernel of the archiving system, very important is the possibility of representing the scene on the basis of objects and background interacting each other (object-based coding). The standard provides also a protocol to create video sequences made up of natural and synthetic objects (SNHC) and of subdivide and transmit the stream in the most suitable way according to channel characteristics (code scalability).<br/>Further reference standards of recent definition, ensure the possibility to implement specific systems functions such as: effective representation of data for the purpose of indexing and retrieval (standard MPEG-7); accurate encoding of specific video frames (or parts) with particular significance with respect to the application, as for instance the image of a human face in a personal identification process (standard JPEG-2000).<br/>Although the standards so far mentioned contain in their specifications all of these features, a high level of freedom is left to the designer in the definition of how to obtain the coded stream and the relevant functions. For instance, at present do not exist any implementation of the MPEG-4 encoder able to achieve a real object-based coding starting from a standard audiovisual stream. This problem is clearly related to the difficult in automatically extracting the objects from the scene in a reliable way, as well as to reconstruct the background from partial/occluded views. Correspondingly, even if the MPEG-7 standard provides a detailed syntax to represent and store the characteristics of a multimedia object, no standard method is yet defined to define which those characteristics are, and how to extract them from the original signal.
Keywords MPEG, Image Processing, Multimedia Communications
Fundings 380000 €
Partners
  • UNIVERSITA' DEGLI STUDI DI GENOVA
  • DIPARTIMENTO DI INFORMATICA E TELECOMUNICAZIONI - UNIVERSITA' DEGLI STUDI DI TRENTO
  • UNIVERSITA' DEGLI STUDI DI CAGLIARI
  • POLITECNICO DI MILANO
  • UNIVERSITA' DEGLI STUDI DI UDINE
  • UNIVERSITA' DEGLI STUDI DI BOLOGNA

DISI Sub-project details

Project astract In particular, the research unit at DIT-UniTN will develop the following actions:<br/>- requirement analysis and definition of specifications of the compression and archiving modules, as well as definition of the testing and validation methodologies to be applied to the algorithms and their software implementations (simulation tools, prototypes)<br/>- Compression and transmission of video streams over wireless channels<br/>- Object-Based Video Compression, including the detection and representation of Regions of Interest (RoIs)<br/>- Indexing and retrieval of images<br/>
Keywords Image Compression, Video Streaming, Indexing
Fundings 61900 €
Manager Francesco De Natale