Dense motion tracking mechanism

Abstract

An apparatus to facilitate motion tracking is disclosed. The apparatus includes one or more processors to receive video data including a plurality of point clouds in a time series, convert each of the plurality of point clouds into a deformable model representation and generate an inter-frame mapping between the deformable model representations to track motion between the plurality of point clouds.

1 - 20 . (canceled) 
     
     
         21 . An apparatus to facilitate motion tracking, comprising:
 one or more processors to:
 generate an isosurface mesh for each point cloud in a plurality of point clouds; 
 deconstruct the isosurface mesh into a mesh constellation having a plurality of independent mesh surface elements, wherein a centroid of each surface element comprises a separate point cloud; 
 perform a skinning and articulation operation to assign bone weights to each of the plurality of independent surface elements; and 
 generate an inter-frame mapping to track motion between the plurality of point clouds. 
   
     
     
         22 . The apparatus of  claim 21 , wherein the bone weights describe a contribution percentage of each bone to a movement of plurality of independent surface elements. 
     
     
         23 . The apparatus of  claim 22 , wherein performing the skinning and articulation operation further comprises transferring the bone weights to the centroids in the mesh constellation. 
     
     
         24 . The apparatus of  claim 23 , wherein the one or more processors further determine an elastic model deformation that approximates advection of particles. 
     
     
         25 . The apparatus of  claim 24 , wherein the final inter-frame mapping comprises a list of integers, wherein each integer identifies a corresponding particle in a previous frame. 
     
     
         26 . The apparatus of  claim 25 , wherein the one or more processors perform a re-sample the deformable model to add missing surface and remove created surface sections. 
     
     
         27 . The apparatus of  claim 21 , wherein the one or more processors detect whether a point cloud includes a plurality of characters in a scene and separate each of the plurality of characters. 
     
     
         28 . At least one non-transitory computer readable medium having instructions stored thereon, which when executed by one or more processors, cause the processors to:
 generate an isosurface mesh for each point cloud in a plurality of point clouds;   deconstruct the isosurface mesh into a mesh constellation having a plurality of independent mesh surface elements, wherein a centroid of each surface element comprises a separate point cloud;   perform a skinning and articulation operation to assign bone weights to each of the plurality of independent surface elements; and   generate an inter-frame mapping to track motion between the plurality of point clouds.   
     
     
         29 . The computer readable medium of  claim 28 , wherein the bone weights describe a contribution percentage of each bone to a movement of plurality of independent surface elements. 
     
     
         30 . The computer readable medium of  claim 29 , wherein performing the skinning and articulation operation further comprises transferring the bone weights to the centroids in the mesh constellation. 
     
     
         31 . The computer readable medium of  claim 30 , having instructions stored thereon, which when executed by one or more processors, further cause the processors to determine an elastic model deformation that approximates advection of particles. 
     
     
         32 . The computer readable medium of  claim 31 , wherein the final inter-frame mapping comprises a list of integers, wherein each integer identifies a corresponding particle in the previous frame. 
     
     
         33 . The computer readable medium of  claim 32 , having instructions stored thereon, which when executed by one or more processors, further cause the processors perform a re-sample the deformable model to add missing surface and remove created surface sections. 
     
     
         34 . The computer readable medium of  claim 28 , having instructions stored thereon, which when executed by one or more processors, further cause the processors to:
 detect whether a point cloud includes a plurality of characters; and   separate each of the plurality of characters in the point cloud.   
     
     
         35 . A method to facilitate motion tracking, comprising:
 generating an isosurface mesh for each point cloud in a plurality of point clouds;   deconstructing the isosurface mesh into a mesh constellation having a plurality of independent mesh surface elements, wherein a centroid of each surface element comprises a separate point cloud;   performing a skinning and articulation operation to assign bone weights to each of the plurality of independent surface elements; and   generating an inter-frame mapping to track motion between the plurality of point clouds.   
     
     
         36 . The method of  claim 35 , wherein the bone weights describe a contribution percentage of each bone to a movement of plurality of independent surface elements. 
     
     
         37 . The method of  claim 36 , wherein performing the skinning and articulation operation further comprises transferring the bone weights to particles in the mesh constellation. 
     
     
         38 . The method of  claim 37 , further comprising determining an elastic model deformation that approximates advection of particles. 
     
     
         39 . The method of  claim 38 , wherein the final inter-frame mapping comprises a list of integers, wherein each integer identifies a corresponding particle in the previous frame. 
     
     
         40 . The method of  claim 35 , further comprising:
 detecting whether a point cloud includes a plurality of characters in a scene; and   separating each of the plurality of characters.