Premium
Autonomous initialisation of exterior orientation parameters using a collinearity search‐based solution
Author(s) -
Seedahmed Gamal H.
Publication year - 2008
Publication title -
the photogrammetric record
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.638
H-Index - 51
eISSN - 1477-9730
pISSN - 0031-868X
DOI - 10.1111/j.1477-9730.2008.00465.x
Subject(s) - collinearity , mathematics , rotation (mathematics) , orientation (vector space) , norm (philosophy) , euclidean geometry , projection (relational algebra) , position (finance) , euclidean space , space (punctuation) , euclidean distance , algorithm , mathematical analysis , geometry , computer science , political science , law , operating system , finance , economics
As is well known, a collinearity‐based solution for the exterior orientation parameters (EOPs) has to deal with their initial approximation. Although this problem has been thoroughly investigated, contemporary literature continues to report it as a major obstacle. To this end, this paper proposes a general and innovative solution for the EOP approximation problem. This solution is formulated as a computational search for the best hypothesis for the approximate values of the EOPs that explains the known correspondence between the observed image coordinates and their counterparts in object space. In particular, the three ranges of the 3D rotation angles are discretised in a coarse‐to‐fine strategy to generate different hypotheses for their values. These hypotheses are combined to form angular triplets that could serve as candidate solutions for the 3D rotation angles. Then, each angular triplet is substituted in a ‘‘linear version’’ of the collinearity equations to solve directly for a candidate hypothesis of the 3D camera position. Together, the two candidate solutions form a hypothesis for the solution vector of the EOPs. Then each hypothesis of the EOPs performs a back‐projection of the object space coordinates, via the collinearity equations, to compute a hypothesis for the image coordinates. The EOP hypothesis that gives the minimum Euclidean distance, or l 2 norm, between the observed and the computed image coordinates, is selected as the best global solution for the approximate values of the EOPs.