作者: Thomas M. Breuel
DOI:
关键词: Hough transform 、 Artificial intelligence 、 Algorithm 、 Transformation (function) 、 Translation (geometry) 、 Representation (mathematics) 、 Point (geometry) 、 Matching (graph theory) 、 Mathematics 、 Rotation (mathematics) 、 Constraint (information theory) 、 Pattern recognition
摘要: This paper presents a new algorithm in the RAST family of algorithms. algorithms perform geometric matching by exploring intersections between query regions and constraint sets space possible model transformations. are closely related to hierarchical Hough transformations but have more desirable combinatorial properties for object recognition applications. Previous applications were limited need represent explicitly transformation space. introduces methods manipulating arising from image point correspondences implicitly, allowing be applied situations where no simple representation is algebraically possible. The method demonstrated exact bounded error 2D translations rotations.