ProFORMA: Objects in 3D via Webcam
ProFORMA: Pro babilistic F -based O -line R apid M odel A cquisition
Pan Qi, a graduate student at the University of Cambridge, has developed an interesting method for modeling objects using a webcam.
From left to right: (A) Rotate the object with your hand in front of the camera. (B) The formation of a cloud of points obtained from the structure of the motion estimation. (C) Delone triangulation from a cloud of points, the formation of a convex volume of tetrahedra. (D) Coverage mesh derived from possible recursion of tetraidal markup. (E) Texturing the displayed grid surface.
A standard reconstruction of an object relies on the image collection stage and the modeling stage, which takes a long time to verify that the model obtained from the image sequence is accurate.
The author proposes a new system for collecting models, called ProFORMA, which creates 3D models directly from a sequence of images received from a video camera. As the user rotates the object in front of the fixed camera, a partial reconstruction model is displayed in front of the user to help more accurate reconstruction. The program also uses an object tracking system.
The results of the reconstruction of various objects:
Video:
The obtained results are quite accurate and this method can lead to cheaper and significantly accelerated methods for obtaining digital models, and the proposed method works directly without additional collection of information about the object.
Pan Qi, a graduate student at the University of Cambridge, has developed an interesting method for modeling objects using a webcam.
From left to right: (A) Rotate the object with your hand in front of the camera. (B) The formation of a cloud of points obtained from the structure of the motion estimation. (C) Delone triangulation from a cloud of points, the formation of a convex volume of tetrahedra. (D) Coverage mesh derived from possible recursion of tetraidal markup. (E) Texturing the displayed grid surface.
A standard reconstruction of an object relies on the image collection stage and the modeling stage, which takes a long time to verify that the model obtained from the image sequence is accurate.
The author proposes a new system for collecting models, called ProFORMA, which creates 3D models directly from a sequence of images received from a video camera. As the user rotates the object in front of the fixed camera, a partial reconstruction model is displayed in front of the user to help more accurate reconstruction. The program also uses an object tracking system.
The results of the reconstruction of various objects:
Video:
The obtained results are quite accurate and this method can lead to cheaper and significantly accelerated methods for obtaining digital models, and the proposed method works directly without additional collection of information about the object.
')
Source: https://habr.com/ru/post/76375/
All Articles