Optical Flow Estimation: An Error Analysis of Gradient-Based Methods with Local Optimization
IEEE Transactions on Pattern Analysis and Machine Intelligence
SUSAN—A New Approach to Low Level Image Processing
International Journal of Computer Vision
In Defense of the Eight-Point Algorithm
IEEE Transactions on Pattern Analysis and Machine Intelligence
Robust Image Corner Detection Through Curvature Scale Space
IEEE Transactions on Pattern Analysis and Machine Intelligence
N-Dimensional Tensor Voting and Application to Epipolar Geometry Estimation
IEEE Transactions on Pattern Analysis and Machine Intelligence
ASSET-2: Real-Time Motion Segmentation and Shape Tracking
IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.00 |
This paper propose an algorithm by which to achieve robust outlier detection without fitting camera models. This algorithm is applicable for cases in which the outlier rate is over 85%. If the outlier rate of optical flows is over 45%, then discarding outliers with conventional algorithms in real-time applications is very difficult. The proposed algorithm overcomes conventional difficulties by using a three-step algorithm: 1) construct a two-dimensional histogram with two axes having the lengths and directions of the optical flows; 2) sort the number of optical flows in each bin of the two-dimensional histogram in descending order, and remove bins having a lower number of optical flows than the given threshold: 3) increase the resolution of the two-dimensional histogram if the number of optical flows grouped in a specific bin is over 20%, and decrease the resolution if the number of optical flows is less than 10%. This process is repeated until the number of optical flows falls into a range of 10%-20%. The proposed algorithm works well on different kinds of images having many outliers. Experimental results are reported.