An Optimized Method Based on RANSAC for Fundamental Matrix Estimation

Fundamental matrix estimation based on RANSAC will encounter the problems of computational inefficiency and low accuracy when outlier ratio is high. In this paper, an optimized method via modification of the RANSAC algorithm is proposed to solve these problems. First, an isolation forest-based algorithm is performed to detect outliers from putative SIFT correspondences according to distribution consistency of features in location, scale and orientation. Then, a number of obvious outliers are eliminated from putative correspondences, which will enhance the inlier ratio efficiently. Finally, fundamental matrix is estimated with the optimized set. Repeated experiments indicate that the proposed method has testified result in speed and accuracy.

[1]  Philip H. S. Torr,et al.  The Development and Comparison of Robust Methods for Estimating the Fundamental Matrix , 1997, International Journal of Computer Vision.

[2]  Kai Zhang,et al.  A Robust Point-Matching Algorithm for Remote Sensing Image Registration , 2014, IEEE Geoscience and Remote Sensing Letters.

[3]  Matthijs C. Dorst.  Distinctive Image Features from Scale-Invariant Keypoints , 2011 .

[4]  David Nistér,et al.  An efficient solution to the five-point relative pose problem , 2003, 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings..

[5]  Jiri Matas,et al.  Locally Optimized RANSAC , 2003, DAGM-Symposium.

[6]  Jinwen Tian,et al.  Discarding wide baseline mismatches with global and local transformation consistency , 2011 .

[7]  Xavier Armangué,et al.  Overall view regarding fundamental matrix estimation , 2003, Image Vis. Comput..

[8]  Fei Tony Liu,et al.  Isolation-Based Anomaly Detection , 2012, TKDD.

[9]  David G. Lowe,et al.  Fast Approximate Nearest Neighbors with Automatic Algorithm Configuration , 2009, VISAPP.

[10]  Tom Fawcett,et al.  An introduction to ROC analysis , 2006, Pattern Recognit. Lett..

[11]  Richard I. Hartley,et al.  In defence of the 8-point algorithm , 1995, Proceedings of IEEE International Conference on Computer Vision.

[12]  Robert C. Bolles,et al.  Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography , 1981, CACM.