基于预检测的ransac.pdf

RANSAC 1/25 V´aclav Hlav´aˇc Czech Technical University, Faculty of Electrical Engineering Department of Cybernetics, Center for Machine Perception 121 35 Praha 2, Karlovo n´am. 13, Czech Republic hlavac@fel.cvut.cz, http://cmp.felk.cvut.cz LECTURE PLAN Courtesy: O. Chum, J. Matas 1. 2. 3. The RANSAC Algorithm [Fischler, Bolles ’81] 2/25 In: U = {x } set of data points, |U | = N i f (S) : S → p function f computes model parameters p given a sample S from U ρ(p, x ) the cost function for a single data point x Out: p ∗ p ∗ , parameters of the model maximizing the cost function RANSAC Algorithm 3/25 k := 0 Repeat until P{better solution exists} η (a function of C∗ and no. of steps k) k := k + 1 I. Hypothesis (1) select randomly set Sk ⊂ U , |Sk | = m (2) compute parameters p = f (S ) k k II. Verification (3) compute cost Ck = x∈U ρ(p k , x) (4) if C∗ Ck then C∗ := Ck , p ∗ := p k end Example I: Epipolar geometry estimation by RANSAC 4/25 U : a set of correspondences, i.e. pairs of 2D points data points m = 7 sample size f : seven-point algorithm - gives