A Computational Approach to Edge Detection
IEEE Transactions on Pattern Analysis and Machine Intelligence
Edge evaluation using necessary components
CVGIP: Graphical Models and Image Processing
A survey of image registration techniques
ACM Computing Surveys (CSUR)
Robust Visual Method for Assessing the Relative Performance of Edge-Detection Algorithms
IEEE Transactions on Pattern Analysis and Machine Intelligence
Comparison of edge detectors: a methodology and initial study
Computer Vision and Image Understanding
Contextual and non-contextual performance evaluation of edge detectors
Pattern Recognition Letters
IEEE Transactions on Pattern Analysis and Machine Intelligence
Distinctive Image Features from Scale-Invariant Keypoints
International Journal of Computer Vision
Gaussian-based edge-detection methods-a survey
IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Reviews
Binary Image Registration Using Covariant Gaussian Densities
ICIAR '08 Proceedings of the 5th international conference on Image Analysis and Recognition
Parametric estimation of affine deformations of planar shapes
Pattern Recognition
A comparison study of inferences on graphical model for registering surface model to 3D image
MLMI'11 Proceedings of the Second international conference on Machine learning in medical imaging
Automatic detection of the magnitude and spatial location of error in non-rigid registration
WBIR'12 Proceedings of the 5th international conference on Biomedical Image Registration
Recovering projective transformations between binary shapes
ACIVS'12 Proceedings of the 14th international conference on Advanced Concepts for Intelligent Vision Systems
Hi-index | 0.15 |
A new technique is described for the registration of edge-detected images. While an extensive literature exists on the problem of image registration, few of the current approaches include a well-defined measure of the statistical confidence associated with the solution. Such a measure is essential for many autonomous applications, where registration solutions that are dubious (involving poorly focused images or terrain that is obscured by clouds) must be distinguished from those that are reliable (based on clear images of highly structured scenes). The technique developed herein utilizes straightforward edge pixel matching to determine the "best” among a class of candidate translations. A well-established statistical procedure, the McNemar test, is then applied to identify which other candidate solutions are not significantly worse than the best solution. This allows for the construction of confidence regions in the space of the registration parameters. The approach is validated through a simulation study and examples are provided of its application in numerous challenging scenarios. While the algorithm is limited to solving for two-dimensional translations, its use in validating solutions to higher-order (rigid body, affine) transformation problems is demonstrated.