Pulse or digital communications – Bandwidth reduction or expansion – Television or motion video signal
Reexamination Certificate
1999-03-01
2002-09-03
Kelley, Chris (Department: 2613)
Pulse or digital communications
Bandwidth reduction or expansion
Television or motion video signal
Reexamination Certificate
active
06445740
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a method of decoding coded digital signals representative of audiovisual data and available in the form of a continuous bitstream, in view of the binary description of a scene to be rendered on a displaying device, said method comprising a processing operation based on an evolutive syntactic language and provided for extracting from said bitstream, in a first step, distinct elements called objects according to the structure of said scene, defining, in a second step, an individual animation of said elements of the scene, defining, in a third step, particular interactions between a user and said elements, and organizing, in a fourth step, specific relations between said scene elements and corresponding individual animations and/or user interactions according to various classes of applications. This invention will be mainly used in the future MPEG-4 decoders.
2. Description of the Related Art
The most important goal of the well-known MPEG-1 and MPEG-2 standards, dealing with frame-based video and audio, was to make storage and transmission more efficient by compressing the concerned data. The future new MPEG-4 decoding standard will be fundamentally different, as it will represent the audiovisual scenes as a composition of objects rather than pixels only. Each scene is defined as a coded representation of audiovisual objects that have given relations in space and time, whatever the manner in which said given scene has been previously organized in these objects (or segmented).
Up to now, the standardization bodies dealing with natural and synthetic sources used to be different. As good three-dimensional (3D) capabilities are becoming an increasingly important part of many fields, including multimedia and World Wide Web applications that use VRML (VRML—or Virtual-Reality Modelling Language—is now the standard for specifying and delivering 3D-graphics-based interactive virtual environments), the MPEG-4 standard considers jointly the natural materials (video, audio, speech) and the synthetic ones (2D and 3D graphics and synthetic sound) and tries to combine them in a standardized bitstream, in view of the presentation of such a multimedia content on a terminal screen. In order to compose this audiovisual information within the scene, their spatio-temporal relationship needs to be transmitted to the terminal.
The MPEG-4 standard defines a syntactic description language to describe the binary syntax of an audiovisual object's bitstream representation as well as that of the scene description information. More precisely, the MPEG-4 system Verification Model 4.0 proposes, for the description of the scenes, a binary format called the Binary Format for Scenes (BIFS). This description, constructed as a coded hierarchy of nodes with attributes and other information such as event sources and targets, is based on the assumption that the scene structure is transmitted as a parametric description (or a script) rather than as a computer program. The scene description can then evolve over time by using coded scene description updates. The node descriptions, which are conveyed in a BIFS syntax, may also be represented, for the purpose of clarity, in a textual form. Some MPEG-4 nodes and concepts are direct analogues of the VRML 2.0 nodes. Others are modified VRML 2.0, still others are added for specific NPEG-4 requirements. Like the VRML 2.0 syntax, the BIFS has provisions for describing simple behaviors and interaction with the user through an event passing mechanism. However some problems, explained hereunder, are not solved by this format.
The first of these addressed problems concerns an unified description of a mixed 2D and 3D scene. There is indeed a fundamental difference between the description of a purely 3D scene, the description of a purely 2D scene, and the description of a mixel 2D/3D scene. In a 3D scene, the layering of the objects is based on the depth information. In 2D, the notion of depth is absent and the layering should be defined explicitly. Furthermore, mixing 2D and 3D objects may be accomplished in several ways:
(1) embedding of 3D objects in a 2D scene:
(a) this is, for example, the case when one tries to render 3D objects in front of a 2D background: in this case, when the user navigates in the scene, the background does not move;
(b) another example is an application in which the user interface contains 2D objects (such as, buttons or text) and a 3D viewer where the scene is rendered;
(2) embedding of 2D objects in a 3D scene:
(a), this is for example, the case when one uses a video object as a texture map on 3D objects;
(b) another example is a texture made of 2D graphic objects (a special case of this is an “active map”, that is a 2D plane in 3D scene made of several composited 2D objects);
(3) these two schemes may be mixed recursively, for example, for embedding 3D objects in a 2D scene and using the resulting composition as a texture map on 3D objects (this may be used to simulate the reflection of a mirror);
(4) a last possibility is to view simultaneously, the same 3D scene from different view points.
At that moment, it is not possible to describe all these possibilities using a single scene graph. A scene graph is a tree that represents a scene by means of a hierarchy of objects called nodes. The scene is composed of grouping nodes and children nodes. The role of grouping nodes is to define the hierarchy and the spatial organization of the scene. Children nodes are the leaves of the tree. These nodes are used to define geometric objects, light sources as well as various types of sensors (objects that are sensitive to user interaction). Grouping nodes have children nodes. These children may be children nodes or other grouping nodes.
All nodes may have attributes which are called fields. The fields may be of any type. For example, sphere is a geometry node. It has a field that defines its radius. It is a single value field of type float (SFFloat). Children nodes of a grouping node are specified in a special field. This field is a multiple value field (a list of nodes), and each value is of type node (MFNode).
Now, in order to define animations and user interaction in the scene, it is possible to make connection between fields using an event passing mechanism called routing. Routing a field A to a field B means that whenever field A changes, field B will take the same value as field A. Only fields of the same type (or the same kind) may be connected. Fields may be specialized: some may only be the destination of a route, they are called eventln, others may only be at the origin of a route, they are called eventOut, others may be both the origin and destination of routes, they are called exposedField and, at last, others may not be connected, they are simply called field.
In VRML, four nodes (Viewpoint, Background, Fog and NavigationInfo) play a special role in the sense that only one of each may be active at a given time. These nodes are said to be bindable nodes.
There are many reasons to try to integrate both 2D and 3D features in one coherent framework:
it is possible to use the same event passing mechanism for the whole 2D/3D scene;
the representation of content can be more compact;
the implementation can be optimized because 2D and 3D specifications have been designed to work together.
In order to fulfill these requirements, one needs to be able to compose, in a 2D space, 2D and 3D layers representing the result of the rendering of a 2D or a 3D scene, as well as using the result of rendering of a 2D or 3D scene as an input to other nodes in the scene graph.
Other problems, not still solved, have also to be considered, especially the following ones:
(1) interactivity with the 2D objects: it may be necessary to be able to interact with the objects, change the layering, add or remove objects, which is not possible without a method to set the depth of a 2D object that is compatible with the event passing mechanism of VRML 2.0;
(2) single event routing mechanism, in order to
Mallart Raoul
Signes Julien
Bugg, Jr. George A
Gross Russell
Kelley Chris
Koninklijke Philips Electronics , N.V.
LandOfFree
Audiovisual data decoding method does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Audiovisual data decoding method, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Audiovisual data decoding method will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2828704