Authors:

Description:

We have decided to modify an existing method to scan objects in 3D developed by Jean-Yves Bouguet. This method consists of projecting a moving shadow line on an object in order to extract its 3D properties.
There are quite a few unfortunate issues associated with projecting a shadow line:

1. A light source is necessary in order to create the shadow.
2. The shadows due to the shape of the object, even though they might be on the path of the projected line, will inevitably result in holes in the meshing, even though the camera can fully see these areas.
3. The projected line cannot be focused on the object. This results in not being able to generate sharp boundaries, resulting in significant errors in the reconstruction process.

1. The laser acts as its own light source. Since it is very compact, it can easily be moved around, allowing us to scan the object from different views while not moving the camera or object (ie. making pixel to pixel correspondence possible). This was previously possible only when using a small light source such as a desk lamp, excluding outdoor experimentation.
2. The laser illuminates the parts of the object located in dark areas. It might even prove to be helpful to scan in a mildly lit environment. A completely dark room may create too much contrast and saturate the CCD. Also, it is now possible to scan objects regardless of color or design.
3. We have chosen a focusable laser (ALL3-3.5G-650 FOV 60° from World Star Tech) which allows the projection of very sharp edges. Since it has a wide fan of view (60°), we can scan only using the center of the beam, thus we don't have to worry about differences in the projected intensity due to the gaussian distribution of light along the beam.

Milestone:

• Week 1:
  Familiarize with the existing method using shadows. Experiment generating meshes for several objects. These meshes will be used to compare results later on.
  
• Week 2:
  Adapt the method substituting the shadow edge with a laser beam. this includes being able to recognize the moving laser on the pictures.
  
• Week 3:
  Take several sets of data on different objects in order to estimate the capabilities and limitations of this method. Compare the new results with the ones obtained with the shadows.
  
• Week 4:
  Preparation of a demo and final presentation of the project.
  

• Presentation given June 5, 1998
• Meshes available

A few pictures:

Laser on angel

Scanning in the dark

Scanning

Page written May 5, 1998 by Christophe Basset <basset@cco.caltech.edu> and Lori Robison <robison@cco.caltech.edu>