DAISY: An Efficient Dense Descriptor Applied for Wide Baseline Stereo


We show that it is possible to estimate depth from two wide baseline images using a dense descriptor. Our local descriptor, called DAISY, is very fast and efficient to compute. It depends on histograms of gradients like SIFT and GLOH but uses a Gaussian weighting and circularly symmetrical kernel. This gives us our speed and efficiency for dense computations. We compute 200-length descriptors for every pixel in an 800x600 image in less than 5 seconds.

This research was conducted at EPFL - Computer Vision Laboratory in collaboration with Vincent Lepetit and Pascal Fua and has also been used for multiview stereo reconstruction

Updated: Thursday, July 24, 2014 10:29:21 +0200