International Business Machines Corporation

Institut Fuer Rundfunktechnik Gmbh

Joanneum Research Forschungsgesellschaft Mbh Institute of Information Systems

Portugal Telecom Inovacao, Sa

Institut Fur Rundfunktechnik Gmbh

Bernard Merialdo

Gerhard Stoll

Benoit Huet

Jörg Deigmöller

Joerg Deigmoeller

Helmut Neuschmied

Andreas Kriechbaum

Jose Bernardo Dos Santos Cardoso

Fausto Jose Oliveira De Carvalho

Roger Salgado De Alem

Remi Trichet

Fausto José Oliveira De Carvalho

José Bernardo Dos Santos Cardoso

To improve a cropping system by obtaining coverage of a wide range of contents for smaller sized displays of handheld devices, a method starts from a metadata aggregation and corresponding video, e.g. in post-production, program exchange and archiving, wherein (a) video is passed to a video analysis to deliver video, e.g. by use of motion detection, morphology filters, edge detection, etc., (b) separated video and metadata are combined to extract important features in a context wherein important information from the metadata is categorized and used to initialize a dynamically fitted chain of feature extraction steps adapted to the delivered video content, (c) extracted important features are combined to define regions of interest (ROI) which are searched in consecutive video frames by object tracking, the object tracking identifies a new position and deformation of each initialized ROI in consecutive video frames and returns this information to the feature extraction thereby obtaining a permanent communication between the feature extraction and the object tracking, (d) one or plural ROIs are extracted and input video frame by video frame into a cropping step (e); based on weighting information a well composed image part is cropped by classifying the supplied ROIs by importance, and (f) the cropped image area(s) are scaled to the desired small screen size.

Method of adapting video images to small screen sizes

A system architecture for providing human intelligible information by processing a flow of input data; e.g., converting speech (source information) into printable data (target information) based on target-dependent probabilistic models; and for enabling efficient switching from one target field of information into another. To that end, the system is provided with a language modeling device including a data base loadable with an application-dependent corpus of words and/or symbols through a workstation; and a language modeling processor programmed to refresh, in practice, a tree-organized model, efficiently, with no blocking situations, and at a reasonable cost.

Method and apparatus for adaptively generating field of application

Growing community of inventors

Valbonne, France

Bernard Merialdo