In image processing is possible to detect a corner by looking at the eigenvalues of the Harris matrix.
Harris matrix can be here http://docs.opencv.org/doc/tutorials/features2d/trackingmotion/harris_detector/harris_detector.html A corner is detected if there is a big variation in both the x and y direction, but how can this be translated to use eigenvector and eigenvalues. How just same outcome occur when the Eigenvalues is big?