Fast iterative five point relative pose estimation

Abstract

Robust estimation of the relative pose between two cameras is a fundamental part of Structure and Motion methods. For calibrated cameras, the five point method together with a robust estimator such as RANSAC gives the best result in most cases. The current state-of-the-art method for solving the relative pose problem from five points is due to Nistér [9], because it is faster than other methods and in the RANSAC scheme one can improve precision by increasing the number of iterations.In this paper, we propose a new iterative method, which is based on Powell's Dog Leg algorithm. The new method has the same precision and is approximately twice as fast as Nistér's algorithm. The proposed method is easily extended to more than five points while retaining a efficient error metrics. This makes it also very suitable as an refinement step.The proposed algorithm is systematically evaluated on three types of datasets with known ground truth.

Robust Three-View Triangulation Done Fast

“…Otherwise, compute the gradient g = J T r and the steepest descent step h SD = −αg (for the computation of the step length α, see [14]). If the SD step leads outside the trust region (α g 2 ≥ Δ), a step in the direction of steepest descent with length Δ is applied…”

“…In all three cases, the new parameter vector is only used in the next iteration (w = w new ) if the gain factor ρ is positive (for the computation of ρ see [14]). Depending on the gain factor, the region of trust is growing or shrinking.…”

Multi-robot relative localisation using computer vision

“…In their work, they used a feature-based method that combines 3D-to-2D camera pose estimation and RANSAC for outlier rejection. Nister also proposed a minimal solver for five points with a motion hypothesis in RANSAC [43], which became popular and started to be used as a benchmark in various works such as [44][45][46][47]. Tardif et.…”