This page contains the Matlab implementation of the quaternion 5-point algorithm (together with the 6-poinr and 7-point versions) based on the relinearization method. RANSAC implementation of the algorithm tested on the TUM dataset is further included.

"Demo_Ver1_1.m":

This Matlab script shows how the relinearzation-based quaternion 5-point algorithm can be called to estimate the relative pose from 5 matched feature points in 2 images.

"Ransac_Tum_V2_1.m":

Run this script to see how the RANSAC implementation of the relinearzation-based quaternion 5-point algorithm can be used to estimate the realtive pose between keyframes in the TUM image dataset (https://vision.in.tum.de/data/datasets/rgbd-dataset).

Matlab scripts are written and tested in MATLAB R2017b, 64-bit Windows OS.

You can download the files individually from the Google Drive folder (below), or you can download all files in the .zip format directly from here.