/2D_3D

A computer vision project using a series of photographs taken using a turntable to generate a 3D point cloud through SIFT and point triangulation.

Primary LanguagePython

2D_3D Reconstruction

Project Webpage: https://dionysus.works/project/2d_3d/

Team: Dionysus Cho + Raag Kashyap

Process

  • preprocess images
    • calibrate camera parameters
    • calculate camera positions
    • open fold of images
    • undistort images
  • keypoint detection
  • feature matching between images
  • triangulation & color
  • point cloud generation

How to Use:

Simply run main.py.

The important lines are:

34> cloud_pts, cloud_rgb = process_img_folder(folder, loop)

which calls process_img_folder on a set of images (the 360 sequence of the Toy from the Turntable) saved in the imgs folder.

39> save_point_cloud(cloud_pts, cloud_rgb, "prinplup")

which saves the processed images from line 34 to the output folder.

40> plot_point_cloud(cloud_pts, cloud_rgb)

which plots the point cloud from line 34 in the browser