Direct 3-D drawing by employing camera view constraints

摘要

An interactive 3-D drawing method supports 3-D modeling of real-world scenes captured in the form of multiple images taken from different locations and angles. The method enables the user manipulate a 3-D drawing primitive without changing its appearance on a selected input image.

主权项

1. A computer-implemented method comprising:
providing a 3-D drawing based on 2-D images of a scene taken from multiple different views, wherein the 3-D drawing has a plurality of 3-D drawing primitives, and the 2-D images comprise at least a first 2-D image comprising a first point of the scene from a first view and a second 2-D image comprising the first point of the scene from a second view, different from the first view; permitting the 3-D drawing primitives to be drawn on a computer screen as a polygonal mesh and the polygonal mesh is overlaid on the 2-D images; using a computer, allowing a user to manipulate a 3-D drawing primitive in a view-locked mode, wherein in the view-locked mode, altering of an appearance of the 3-D drawing primitive occurs without changing its appearance on a selected input image; allowing the user to move the 3-D drawing primitive in 3-D under a locked constraint to match its appearance to corresponding features on other 2-D images; allowing view-locking of the 3-D drawing primitive on one 2-D image while aligning on the another 2-D image; allowing moving of the 3-D drawing primitive on a 3-D plane parallel to a current viewing camera to match its appearance to the corresponding features on other images; and allowing display on the computer screen a surface corresponding to the polygonal mesh of the 3-D drawing primitive, without a 2-D image overlay.