Method and apparatus for parallel multi-view point video capturing and compression
AssigneeReality Commerce Corporation
InventorPing Liu
Life of this patent
Estimating the term…
Abstract
A method and apparatus for digital video capturing and compression supports multi-viewpoint interactive video applications. Multiple (typically hundreds of) video cameras centripetally placed in a three dimensional space acquire real-time visual pictures of the same 3D scene from multiple viewpoints at the rate of 20-30 frames per second (fps). Parallel Multi-Viewpoint Video Capturing and Compression (PMVCC) supports the operation of the apparatus. In an exemplary realization, a Multi-DSP Array (MDA) architecture uses hundreds of high-performance digital signal processors (DSPs) working in parallel to obtain a sustained computation power of hundreds of GIPS (Giga Instructions Per Second).
Claims24
An integrated digital video apparatus comprising:
- a cover with a plurality of digital video cameras located thereon;
- said cameras being grouped into a plurality of processing groups, each of said cameras belonging to one of said processing groups;
- said processing groups including a control camera and a one or more surrounding cameras;
- digital signal processors corresponding to said cameras;
- said control camera for each said processing group providing video representing an independent viewpoint, said one or more surrounding cameras in each processing group providing a corresponding number of dependent viewpoints surrounding said independent viewpoint;
- and
- a controller receiving the output of said digital signal processors for one of said processing groups, and applying a video coding scheme to only said independent viewpoint to provide a coded central image, and deriving prediction data on the images of said dependent viewpoints based on said coded central image.
A method of video capture and compression, comprising:
- providing a cover comprising a plurality of digital video cameras located thereon, each said camera being located at a defined viewpoint;
- grouping said cameras into at least two processing groups, each said processing group having only one of said cameras as a control camera and the remainder of said cameras of said processing group being surrounding cameras;
- providing digital signal processors corresponding to said cameras, wherein said surrounding cameras provide output to only a corresponding one of said digital signal processors, and said control camera provides output to said corresponding digital signal processors and also to said ones of said digital signal processors corresponding to said surrounding cameras;
- providing respective digital video data from said cameras as said output to said digital signal processors, said output from said control camera representing an independent viewpoint, said surrounding cameras in each said processing group providing a corresponding number of dependent viewpoints surrounding said independent viewpoint;
- and
- applying a video coding scheme to only said independent viewpoint to provide a coded central image for said corresponding processing group, and deriving prediction data on the images of said dependent viewpoints based on said coded central image.
Dependencies are inferred from the claim text.