Speeded up robust feature detection and matching

SURF approximates Laplacian of Gaussian with computation based on a simple two-dimensional box filter as described in the last chapter.  The convolution with the box filter can be easily calculated with the help of integral images, which improves the performance of the algorithm. SURF relies on the determinant of the Hessian matrix for both scale and location. The approximated determinant of Hessian can be expressed as:

Where w is a relative weight for the filter response and used to balance the expression for the determinant. The Dx, Dy are the result of the Laplacian operator in X- and Y

Get Hands-On GPU-Accelerated Computer Vision with OpenCV and CUDA now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.