Structured Lighting

COMPSCI 775ST Assignment 5

From:04/10/2002
Due: 18/10/02 2:00pm

Aim

Resource

Goal

Details

Hints

Questions to be answered

• Understand Structured lighting with single light plane approach
• Understand 3D reconstruction using structured lighting

Four sets of images are available for this assignment.  You also need the images for the calibration plane. The image have been zipped into the following files.
 

• calibplane- images of the calibration plane.
• box- images of a box
• bunny- images of a bunny
• teapot- images of a teapot
• teapotr- images of teapot rotated with respect to another axis

The following table contains the Z value of the the calibration plane.

calibplane1.bmp	-120 mm
calibplane2.bmp	-84 mm
calibplane3.bmp	-48 mm
calibplane4.bmp	-12 mm
calibplane5.bmp	+24 mm

Each image series contains 36 pgm images. There is a 10 degree rotation with respect to the Yw axis between successive images.

The rotation axis is projected on image column 128.

Note: The light profile in teapotr*.pgm is not visible in several of the images.

• Determine the function which transforms the light profile column values into depth values
• Reconstruct the object(s) from the chosen set(s) of images.
• Visualize the object(s).

• Explain Structured lighting, especially with respect to the single light plane approach.
• Use calibplane*.pgm to construct a look-up-table (LUT). Note that only5 points are given by the calibration plane, therefore, these points have to be interpolated.
• Explain your approach in preparing the LUT, extracting the light profile and calculating the depth values.
• Choose your set(s) of images.
• Extract the light profile for each image.
• Calculate the depth of the object according to the horizontal displacement of the light profiles.
• Reconstruct the object using cylindrical co-ordinate system.
• Visualize the reconstructed object.
• Comment on your results.

Calibration

Please have a look at page 140-142 of the lecture notes.

Since the width of the light profile is not exactly 1 pixel, you might need to model the distribution of the light profile before extracting the light profile.

Reconstruction of the object

Please have a look at page 148 and 152 of the lecture notes.

It is recommended that you use cylindrical co-ordinate system.

Visualising the results

You may represent the boundary of a cross-section of the objectas a 36-sided polygon. Or better, you may use VRML to visualise the object as a 3D model.

Have a look at http://www.cs.auckland.ac.nz/references/#html for information on visualisations packages.

• What assumptions have you made?
• What are the possible sources of error in Structured lighting?

This page was orignally maintained by Qi
Now it's a local copy.