针对传统的ORB(Oriented fast and rotated brief)算法在运算速度以及精度方面有时难以满足某些应用场合实际要求，在特征点提取阶段，利用金字塔光流法提取特征点并划分有效及无效区域特征点，从而降低特征点匹配个数和提高后续运算特征点匹配速度;在特征点匹配阶段，将传统算法中的欧氏距离改为曼哈顿距离，再用 MLESAC 算法来剔除误匹配点。将 SURF(Speeded up robust features)算法、SIFT(Scale-invariant feature transform)算法、ORB算法和改进后的ORB算法对光照条件不同、模糊度不同以及尺度大小不同的两张图像进行处理，改进后的ORB算法无论是在匹配速度还是匹配精度方面相比于传统ORB算法都有了明显改善。
Image Processing Method Based on Improved ORB Features
As for the traditional ORB(Oriented Fast and Rotated Brief)algorithm,it is sometimes difficult to meet the actual requirements of certain applications in terms of computing speed and accuracy.In the feature point extraction stage,the pyramid optical flow method is used to extract feature points and divide the effective and ineffective regions into fea-ture points,so as to reduce the number of feature point matches and improve the speed of feature point matching for the subsequent operations.In the feature point matching stage,the Euclidean distance in the traditional algorithm is changed to Manhattan distance,and finally the MLESAC algorithm is used to eliminate the false matching points.The SURF(Speeded up robust features)algorithm,SIFT(Scale-invariant feature transform)algorithm,ORB algorithm and the improved ORB algo-rithm are used to process two images with different lighting conditions,blurring degrees and scale sizes.The improved ORB algorithm is superior to the traditional ORB algorithm both in terms of matching speed and matching accuracy.
information entropyManhattan distancemaximum likelihood consensus
国家科技期刊平台
NSTL
万方数据
