A Weighted Total Least Squares Adjustment Method for Building Regularization

An errors-in-variable (EIV) with arbitrary constraints is proposed for the purpose of building regularization from remote sensed data, in which the edge points are treated as measurements, the constrained slopes and intercepts of each edge are chosen as parameters. Assuming the measurement vector and the design matrix are mutually correlated, the scheme of calculating the dispersion matrix of augmented matrix is suggested. A generic constrained weighted total least squares(WTLS) algorithm is derived with an approximate accuracy assessment method, and the WTLS algorithm of a quadratic constrained EIV problem is given as a specific case. Theoretic analysis and data experiment demonstrate the advantages of an EIV model compared with a Gauss-Helmert model in building regularization problem, and the rapid convergence rate of proposed WTLS algorithm. It aims to promote WTLS adjustment methods, and to expand the applications of total least squares method in new surveying technology with a certain theoretical and practical significance.