Premium
Progressive registration of image features and 3D vector lines for orientation modelling
Author(s) -
Chang WenChi,
Hung ChingHui,
Chen LiangChien
Publication year - 2018
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/phor.12227
Subject(s) - collinearity , artificial intelligence , computer vision , orientation (vector space) , computer science , pixel , line (geometry) , feature (linguistics) , image (mathematics) , georeference , pattern recognition (psychology) , mathematics , geography , geometry , linguistics , philosophy , physical geography
This paper proposes a progressive method to register image features and three‐dimensional (3D) vector lines for image orientation modelling. Directed at images acquired with an approximate direct‐georeferencing capability, this study employed 3D vector lines extracted from a geographical information system ( GIS ) database as ground control. The major work comprises three steps: (1) straight lines were parameterised and applied as control in the collinearity condition equations to determine exterior orientation parameters ( EOP s); (2) coarse registration employed quadrangle features formed from straight lines to modify the initial EOP s; and (3) a two‐step fine registration, initially involving all line‐feature candidates, followed by the selection of the most probable one from neighbouring lines. Experimental results indicate that the proposed methods can achieve an accuracy of about 2 pixels (close to the accuracy of the employed GIS database) from initial EOP s with more than 800 pixel errors. Accuracy improvements in each step of the proposed coarse‐to‐fine registration are also demonstrated.