Abstract
Several estimation problems in vision involve the minimiza- tion of cumulative geometric error using non-linear least-squares fit- ting. Typically, this error is characterized by the lack of interdependence among certain subgroups of the parameters to be estimated, which leads to minimization problems possessing a sparse structure. Taking advan- tage of this sparseness during minimization is known to achieve enormous computational savings. Nevertheless, since the underlying sparsity pat- tern is problem-dependent, its exploitation for a particular estimation problem requires non-trivial implementation effort, which often discour- ages its pursuance in practice. Based on recent developments in sparse linear solvers, this paper provides an overview of sparseLM, a general- purpose software package for sparse non-linear least squares that can exhibit arbitrary sparseness and presents results from its application to important sparse estimation problems in geometric vision.