Increasing the Efficiency of 3D-geometry Extraction Using Semantic Analysis of Images

M. G. Kazakov , e-mail: mike.kazakov@gmail.com

A method for increasing the efficiency of 3D-geometry extraction is proposed. This method is based on semantic analysis of images using segmentation techniques and correlating the segments with affine transformations. Method details are given with according visualization and constants values are shown with graphs.

Keywords: computer vision, structure from motion, 3D-reconstruction, image segmentation, affine transformations, interest point descriptors

pp. 42–47