Depth Estimation from Stereoscopic Image Pairs Assuming Piecewise Continuos Surfaces

3D Reconstruction of Remote Sensing Image Using Region Growing Combining with CMVS-PMVS

“…(a) Determination the extension of cell: for patch p, collect cell image of neighbors in each visible image to initialize C(p): (4) If a patch has been rebuilt, it is not necessary to expand. Specifically, if an image cell has a neighbor patch to p, remove from C(p).…”

“…The local methods such as gradient descent, level set, or global method such as image segmentation, have high reconstruction precision. But the process of rebuilding demands extra information, such as depth maps, bounding box, visual shell and so on, which are not suitable for the reconstruction of chaotic image sets and outdoor scenes [3][4][5].…”

Reading from Scratch – A Vision-System for Reading Data on Micro-structured Surfaces

“…Both types use knowledge about the set-up geometry. Depth from stereo [7] builds a depth map of the surface using a stereo microscope observing the same image from different perspectives. Using the pixel-wise found disparity along the epipolar line and the knowledge about the relation of both cameras, the depth map can be reconstructed.…”

A hierarchical stereo algorithm using dynamic programming

“…First, it is compared with the algorithm of Koch [12] and Falkenhagen [10]. They developed an algorithm that searches for correspondences by matching blocks (for instance NCC).…”

“…A second alternative is a more complex dissimilarity function that implements a cross correlation between the neighbourhoods of the pixels under consideration [10]. It is implemented according to following formula:…”