Robust recognition of buildings in compressed large aerial scenes

Accurate Building Structure Recovery from High Resolution Aerial Imagery

“…So we define a global classification in comparison with a mean ground elevation in the scene, which is computed from the region of lowest elevation. Actually, it means that ground is considered as flat on each image; this hypothesis is not too restrictive and it is easy to extend the method by dividing large images into subimages before extracting the lowest regions 3 . Regions which have a mean elevation of a certain threshold T a larger than this mean ground elevation are labeled as above-ground.…”

“…To build a pyramid of gradient pictures, we use a similar method to the one described in [3]. First, the initial image is smoothed and sampled to obtain the reduced images.…”

Building Detection and Reconstruction from Mid- and High-Resolution Aerial Imagery

“…To build the pyramid of gradients, we use a method similar to the one described in [1]. First, the initial image is smoothed and subsampled to obtain the reduced images.…”

Hausdorff Distances for Searching in Binary Text Images

“…A parameter α = K/N A , (0 < α ≤ 1) can be defined, relative to the set A. D.-G. Sim et al [16] claim that α ≈ 0.4 provides good matching image results. The idea of R. Azencott et al [5], and J. Paumard [15] is that we do not take into account the L closest neighbors of a ∈ A in B. So we define the distance from a point a ∈ A to the set B as follows…”