ECE 564 Computer Vision Assignment 8

Submit HTML documentation and MATLAB code on Isidore. See stereo3 for some general hints.

  1. Use 2019-03-04_17-26-54.JPG (shown below) to construct a directed homography as demonstrated in class showing the hallway as would be seen from directly above (parallel hall walls)

  2. Find the camera poses for product groups 2 and 3. Use group1 for guidance. Compare the three groups.

  3. Use an image set (two images) of your product. Use cpselect to choose matching points from each image. Use estimateFundamentalMatrix to calculate the Fundamental matrix and verify that img2*F*img1' = 0 for matching image points. Calculate the Essential matrix from the fundamental matrix, using your intrinsic matrix from an earlier calibration.

  4. Use the same image set and the extrinsics from the second problem to find the Essential matrix. Verify that x2*E*x1' = 0 for matching normalized camera points. Then calculate the Fundamental matrix from the Essential matrix. Compare to the fundamental matrix calculated in the previous problem.

  5. Use the extrinsics obtained in the previous problem to project matching points back to world coordinates. Compare the results to known dimensions of your product.

  6. Use the same image set and fundamental matrices deterined earlier. Select a feature in image 1 and its corresponding image point (img1) to determine the epipolar line in image 2. Draw the line on image 2 and verify that it passes through (or very close to) the matching point in image 2. Draw two such epipolar lines and calculate their intersection. Show that this point (epipole) can be obtained from null(F).


Maintained by John Loomis, last updated 18 April 2016