A combined solution is proposed for solving the point-plane variational problem in a closed-form based on fusion visual features and 3d point clouds. An accurate method for reconstructing 3d scene is introduced, and the closed-form solutions of registration task for orthogonal transformation are presented. This method is used to reconstruct a threedimensional model of the environment from a set of images and depth map. The given method is used to match and register point clouds with an arbitrary spatial resolution and an arbitrary scale with respect to each other. The suggested method improves the convergence and accuracy of reconstruction methods for dynamic, bulky scenes.

This The scientific problem at solving which the present project is directed consists in the development of accurate methods for reconstruction of a three-dimensional map of the accessible of environment with requied accuracy of reconstruction. The problem of consistent aligning of 3D point data is known registration task. The most popular registration algorithm is the Iterative Closest Point algorithm. Three basic problems are characteristic for the ICP algorithm: first, the convergence of the algorithm depends strongly on the choice of the initial approximation; second, the algorithm does not take into account the local shape of the surface around each point; and, third, the search for the nearest points is of high computational complexity. In this paper a new close solutions to 3D total variation regularization will be obtained and effective algorithms for restoring 3D data will be designed. The proposed approach improves the accuracy and convergence of reconstruction methods for complex and large-scale scenes. The performance and computational complexity of the proposed RGB-D Mapping algorithm in real indoor environments is discussed.