An Improved ORB Algorithm of Extracting and Matching Features

Abstract

For feature extraction of image mosaicing, an improved fast extracting algorithm of binary feature points is presented which is based on ORB (Oriented FAST and Rotated BRIEF). In the process of detection using median filter method to detect more accurate feature points. Improving the speed by using the ORB algorithm to extract image binary descriptors and using RANSAC algorithm and homography matrix to eliminate false matching points. In the end, making use of perspective transform to map points for obtaining the target coordinates to correct images and achieve two images matching in the scene. The experiment is under the environment of OpenCV and Visual Studio2010.The result shows that the improved ORB algorithm is faster and more accurate.