Correspondence Analysis for Binocular Stereo

(Shirai's Algorithm)

COMPCI 775ST Assignment 2
From: 09/08/2002
Due: 21/08/2002 2:00pm

• Aim

• Resource

• Goal

• Details

• Questions to be answered

• Example of results

• Understand Binocular Stereo
• Understand Shirai's Algorithm
• Know more about Depth Analysis

There are a lot of image files available.  You may choose one pair or more from the following images

• complex_l.bmp and complex_r.bmp
• pyrtasse_l.bmp and pyrtasse_r.bmp
• dia_l_r.bmp, dia_l_g.bmp, dia_l_b.bmp and dia_r_r.bmp, dia_r_g.bmp, dia_r_b.bmp
• macbeth_l_r.bmp, macbeth_l_g.bmp, macbeth_l_b.bmp and macbeth_r_r.bmp, macbeth_r_g.bmp, macbeth_r_b.bmp
• tuch_l_r.bmp, tuch_l_g.bmp, tuch_l_b.bmp and tuch_r_r.bmp, tuch_r_g.bmp, tuch_r_b.bmp

• The base distances of these images are not given.  Please make your own assumptions.
• The images showing views of Auckland are not aligned vertically.

You may use one or more pair(s) of colour images to calculate and to represent the disparities of the pixels pairs. Use Shirai's algorithm for correspondence analysis.

• Describe the Binocular Stereo Model.
• Preprocess the pair of images by finding edge maps.  You may use the results from Assignment 1.
• Implement Shirai's algorithm.
• Perform correspondence analysis on the images using Shirai's algorithm.
• Represent the disparities in a suitable way.
• Verify your results and discuss the errors.
• Compare the results obtained by altering parameters d1, d2 and d3.
• You may apply Shirai's algorithm on three separate colour channels and on the intensity channel to compare the disparities obtained in each case.
• Comment on Shirai's algorithm with respect to your results.

• What assumptions have you made?
• Why do we need to calculate disparities?
• What do the parameters d1, d2 and d3 in Shirai's algorithm represent?