# Differences

This shows you the differences between two versions of the page.

harris_corner_detector [2013/04/14 02:30] jongbor [How to find corners] |
harris_corner_detector [2014/04/19 15:41] (current) |
||
---|---|---|---|

Line 71: | Line 71: | ||

===== Practical matters ===== | ===== Practical matters ===== | ||

- | Since computing eigenvalues can be expensive, it is preferred to calculate the score $R = det(M)-k\cdot trace(M)^2$, where $k$ is an empirically determined constant, in order to compute the "cornerness" of a point. | + | Since computing eigenvalues can be expensive, it is preferred to calculate the score $R = det(M)-k\cdot trace(M)^2$, where $k$ is an empirically determined constant, in order to compute the "cornerness" of a point. $R$ preserves information about the eigenvalues since $det(M) = \lambda_1\lambda_2$ and $trace(M) = \lambda_1 + \lambda_2$ but it is also easier to calculate. |

+ | | ||

+ | MATLAB implementations can be found [[http://www.csse.uwa.edu.au/~pk/research/matlabfns/Spatial/harris.m|here]] and [[http://www.di.ens.fr/willow/events/cvml2013/scholarship/application/harris.m|here]]. |