Due on Feb 29

Please write the Harris corner detector from scratch.

  • Your implementation should include

    • compute the cornerness score of every pixel in the input image

    • threshold the cornerness score to create a mask of corners

    • non-maximum suppression to keep only local maximum of cornerness score

    • Extra credit (+40%): derive a cornerness score in terms of element of \(M = \begin{pmatrix} a & c\\ c & b \end{pmatrix}\) for Shi-Tomashi algorithm. That is, express the cornerness score \(\min\{\lambda_1, \lambda_2\}\) in terms of \(a, b, c\), and \(d\), where \(\lambda_1\), \(\lambda_2\) are eigenvalues of \(M\)

    • Extra credit (+10%): implement the Shi-Tomashi algorithm. You just need to modify the cornerness score.

  • Please test you code on this image.

  • As usual, please submit source code and screenshot(s) of source code and your resulting image.

  • You are free to use either MATLAB or OpenCV/numpy/Python.

  • If you are really stuck, please watch the lecture video again. You can find a solution inside.