An Evaluation of the Performance of RANSAC Algorithmsfor Stereo Camera Calibrarion

Abstract

This paper compares the use of RANSAC for the determination of epipolar geometry for calibrated stereo reconstruction of 3D data with more conventional optimisation schemes. The paper illustrates the poor convergence efficiency of RANSAC which is explained by a theoretical relationship describing its dependency upon the number of model parameters. The need for an a-priori estimate of outlier contamination proportion is also highlighted. A new algorithm is suggested which attempts to make better use of the solutions found during the RANSAC search while giving a convergence criteria which is independent of outlier proportion. Although no significant benefit can be found for the use of RANSAC on the problem of stereo camera calibration estimation. The new algorithm suggests a simple way of improving the efficiency of RANSAC searches which we believe would be of value in a wide range of machine vision problems.

Estimation of Rotational Axis and Attitude Variation of Satellite by Integrated Image Processing

“…The image processing method considered in this study for detecting the rotational axis and attitude variation consists of six steps: (step1) target searching based on color information, (step2) feature points extraction by Harris corner detector (Harris & Stephens , 1988), (step3) optical flow estimation using template matching and random sample consensus (RANSAC) (Lacey et al , 2000), (step4) deleting incorrect paired points using the epipolar condition (Zhang et al , 1995), (step5) initial guess of the rotational axis and attitude variation from the extracted optical flow by a heuristic approach, and (step6) an iterative algorithm for obtaining the precise rotational axis and the attitude variation from the initial guess. In this section, each step will be briefly explained.…”

Spatial Topology Graphs for Feature-Minimal Correspondence

“…While the corners are mostly detected reliably, some corners could be occluded in some images, while others could be T-junctions resulting from occlusion, which change with viewpoints [3]. Features from multiple views are matched using a robust method, such as RANSAC [6], by iteratively computing the fitness of the feature matches to the recovered geometry, and selecting the best fit. This method is expensive in time and does not perform well when the number of correct matches is much lower than mismatches.…”

Unconstrained Segue Navigation for an Immersive Virtual Reality Experience

“…All the matches were finally aggregated for the voting in RANSAC [23] to calculate the homography which was used to warp Image A onto Image B. Using this warp, the four corner coordinates of Image A were calculated after the warp.…”