Method and apparatus for 3D mesh de-noising

A computer implemented method for de-noising a 3D model, comprising:

• accessing a two dimensional image content by a processor configured to process image content information using structural information;
• determining a shortest geodesic path based on said two dimensional image, wherein said shortest geodesic path has multiple mesh edges by said processor to identify a set of mesh edges amongst said multiple mesh edges connecting a first vertex of the 3D model and a second vertex of the 3D model wherein the set of mesh edges is determined responsive to at least one of a distance between the first vertex and the second vertex, curvature variation, and local variation;
• constructing a proximity graph for the first vertex, wherein estimated mesh edges of the proximity graph are used to determine the set of mesh edges;
• determining a respective geometric variation for each one of a plurality of edges, the plurality of edges belonging to the set of mesh edges;
• determining structural information relating to the 3D model in a neighborhood around the first vertex of the 3D model responsive to the respective geometric variations by calculating geometric variations along the shortest geodesic path established based on spatial relationship between the first and the second vertex;
• and
• processing the first vertex in response to the structural information relating to the 3D model.