The Research of SLAM Monocular Vision Based on The Improved SURF Feather
Ye You · 2014
Image feature detection is an important part of monocular vision SLAM system. At present, detection methods can not balance these two requirements of speed and stability. The SURF algorithm in the application of the SLAM, its robustness is slightly worse, sometimes get the less number of correct matching points than usual, aiming at disadvantages of SURF algorithm in SLAM system, this paper combines the SURF descriptor with a fast and effective descriptor-BRIEF. Since the original BRIEF descriptor does not have the rotation and scale invariance, and is sensitive to noise, this paper has also been improved the problem. The Oriented BRIEF-SURF has a stronger matching performance and a better real-time performance, and it is invariant to scale and rotation, reduce the effect of noise.