Open AccessPhotogrammetric stereo image rectification
Author(s) -
Martinus Edwin Tjahjadi,
Fourry Handoko
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1869/1/012066
Subject(s) - computer vision , photogrammetry , artificial intelligence , computer science , point cloud , bundle adjustment , parallax , stereo imaging , process (computing) , computer graphics (images) , structure from motion , point (geometry) , object (grammar) , collinearity , mathematics , motion estimation , geometry , operating system
Today, off-the-shelf digital cameras are hand on tools for conducting visual inspections in surveying and mapping related activities or in structural inspection of industrial objects. Dense point clouds of the inspected object surface are of typical data required and their accuracies are also become more demanding. Deriving these data are off particular interest of this paper instead we work on overlapping images for generating those ones. An imaging geometry of stereo images is exploited further to reveal some drawbacks of the normalized stereo imaging configurations. We utilize a photogrammetric collinearity condition to model physical reality of the imaging process. A sequence of coordinate axes rotations on both images is rearranged as well as preserving normalized images resolutions to be equal to the source ones. Also based upon the prescribed resolutions, an efficient indirect resample is performed to interpolate grey values of the warped images. To ascertain reliability of the mathematically developed method, C++ programming codes were written and implemented in the solution. It reveals that a nuisance such as y-parallax is completely eliminated. Therefore, by using our method, the normalized images can be used in the dense point clouds production in a high quality and efficiency.