Seamless texturing of 3D meshes of objects from multiple views

摘要

Methods and systems for texturing of three-dimensional (3D) object data models are provided. An example method may include receiving information indicating a geometry of an object receiving a plurality of images of the object. The method may also include assigning images of the plurality of images that have a resolution above a threshold to a plurality of polygons that approximate the geometric surface of the object. The method may also include determining adjacent polygons that are assigned to different images of the plurality of images so as to identify boundaries of images and minimizing such boundaries. The method may also include determining a mismatch factor for boundaries of the modified boundaries of images and reassigning images in boundaries having a mismatch factor above a threshold so as to reduce a gradient variation between the images in the modified boundaries.

主权项

1. A method comprising:
receiving information indicating a geometry of an object, wherein the information indicating the geometry of the object comprises a plurality of polygons that approximate at least a portion of a geometric surface of the object; receiving a plurality of images of the object, wherein each of said plurality of images illustrate the object from a different respective angle; assigning images of the plurality of images that have a resolution above a threshold to polygons of the plurality of polygons; determining adjacent polygons that are assigned to different images of the plurality of images so as to identify boundaries of images; determining, by a processor, an image of the plurality of images that is assigned to a threshold number of polygons in a proximity of one of the adjacent polygons; reassigning the image that is assigned to the threshold number of polygons to the adjacent polygons so as to modify the boundaries of images; determining a mismatch factor for boundaries of the modified boundaries of images; and reassigning images in boundaries having a mismatch factor above a threshold so as to reduce a gradient variation between the images in the modified boundaries.