Abstract: A technique for representing a visual scene as a directed acyclic graph of data and operators that generates a sequence of image frames over specified time intervals. The graph specifies temporal and spatial values for associated visual elements of the scene. Time is modeled in the inheritance properties explicitly defined within the scene graph hierarchy, by assigning temporal attributes to each media element. Branch nodes of the graph specify transforms for the temporal and spatial coordinate systems. To evaluate the appearance or behavior of the scene and in particular the global time values of particular elements at a given time instant, the graph is traversed in a direction from a root node down toward the leaf nodes, thereby causing temporal transformations specified along the branches of the graph to modify time parameters of the scene data at the nodes. Child nodes are preferably evaluated after being transformed, to determine the extent to which they contribute the data to the final scene.

Abstract: A technique for temporal smoothing of results from a scene analysis process which creates a sequence of visually pleasing and acceptable images generated in whole or part from such results. The technique applies temporal smoothing across time-related sets of scene analysis results. The spatial smoothing can be applied at various steps in the process: to images in the original sequence, to intermediate or final results of scene analysis in either a pixel-oriented or geometric domain, or to the images generated in whole or part from the scene analysis results. In a preferred embodiment, different levels of smoothing are applied to different parts of the intermediate or final results. The differentiation can be done using image masks (for pixel-oriented results) or geometry selection techniques (for geometric results).

Abstract: A technique for representing a visual scene as a directed acyclic graph of data and operators that generates a sequence of image frames over specified time intervals. The graph specifies temporal and spatial values for associated visual elements of the scene. Time is modeled in the inheritance properties explicitly defined within the scene graph hierarchy, by assigning temporal attributes to each media element. Branch nodes of the graph specify transforms for the temporal and spatial coordinate systems. To evaluate the appearance or behavior of the scene and in particular the global time values of particular elements at a given time instant, the graph is traversed in a direction from a root node down toward the leaf nodes, thereby causing temporal transformations specified along the branches of the graph to modify time parameters of the scene data at the nodes. Child nodes are preferably evaluated after being transformed, to determine the extent to which they contribute the data to the final scene.

Abstract: A technique for displaying a visual representation of an estimated three-dimensional scene structure and the values of various parameters associated with the scene, together with a visual representation of at least one two-dimensional image used in the scene structure estimation algorithm. A user inputs information by adjusting parameters and/or specifying an element or region of the visual representations and supplies mark-ups and other information such as attributes for the element or region to be applied during a next iteration of the scene structure estimation algorithm. The scene structure estimation algorithm is then re-executed and the process repeats until the user is satisfied with the resulting visual scene structure.

Abstract: A method for feature tracking across a plurality of related image frames by developing a dense array of features and their paths as a deformable, regularly spaced grid, with vectors for each nodal feature point in the grid being derived through a process related to optical flow. The resulting dense feature array can be applied to techniques such as tracking two-dimensional shapes across frames, estimating three-dimensional paths of selected feature points, estimating three-dimensional camera paths from multiple feature points, or recovering estimated three-dimensional scene structure (including estimated depths of object surfaces) from feature tracking data.

Abstract: An integrated media production technique for refining a finish quality media content. The technique permits a media production to be defined as a computerized choreography model of a virtual stage containing data objects representing live/recorded image elements as well as synthetic-abstract elements and multiple versions thereof. Elements of the choreography model are iteratively rendered, blended, adjusted, and selected until a desired quality result is achieved.

Abstract: A technique for converging upon a computer-based model of a real world or synthetic scene. The computer model makes use of abstraction-based data objects as well as image-based data objects. A correlation mesh provides links between related image-based and abstraction-based objects. An initial step in a process analyzes an input image stream and user inputs to derive initial image-based objects and abstraction-based objects for the scene model. Subsequent steps in the process allow user inputs to refine the image-based objects, abstraction-based objects and/or the correlation mesh. As a result, refinements to the image-based object model of the scene can improve the abstraction-based model of the scene, and refinements to the abstraction-based object model can improve the image-based model.

Abstract: In a media production system, two-dimensional images captured from physical objects are analyzed to create three-dimensional representations of the physical objects within a virtual stage. Representations of physical and synthetic objects are manipulated and choreographed in the virtual stage within an integrated system which also generates a media product in the form of a two-dimensional image stream of the physical and synthetic objects with optional synchronized audio tracks. A correlation is maintained between object models in the virtual stage and the original two-dimensional images so that information from those images can be included in the final product. To facilitate direction of combined media products, a composite display of the three-dimensional virtual stage and a two-dimensional preview are provided. Also provided are a score display providing temporal representations of elements within the scene and an object catalog of object oriented representations of elements within the scene.