REDUCING STRUCTURE-FROM-MOTION - A GENERAL FRAMEWORK FOR DYNAMIC VISION PART 1 - MODELING

The literature on recursive estimation of structure and motion from mo nocular image sequences comprises a large number of apparently unrelat ed models and estimation techniques. We propose a framework that allow s us to derive and compare all models by following the idea of dynamic al system reduction. The ''natural'' dynamic model, derived from the r igidity constraint and the projection model, is first reduced by expli citly decoupling structure (depth) from motion. Then, implicit decoupl ing techniques are explored, which consist of imposing that some funct ion of the unknown parameters is held constant. By appropriately choos ing such a function, not only can we account for models seen so far in the literature, but we can also derive novel ones.