基于改进SIFT和RANSAC的物体特征提取和匹配的研究  

Research on Object Feature Extraction and Matching Based on Improved SIFT and RANSAC

作　　者：张春林 陈劲杰[1] ZHANG Chunlin;CHEN Jinjie(University of Shanghai for Science and Technoloy,Shanghai 200093,China)

机构地区：[1]上海理工大学,上海200093

出　　处：《软件工程》2018年第11期7-9,6,共4页Software Engineering

摘　　要：使用基于RANSAC提纯的改进SIFT算子和SIFT算子分别在两张不同视角的图像中提取并匹配特征点对,通过实验结果验证了基于改进的SIFT算提取特征点的准确率和匹配的精度都要比SIFT算子和Harris算子高,改进SIFT算子提取匹配出的特征点对更加的有利于目标物体的三维重建工作。The SIFT operator and the improved SIFT operator based on RANSAC purifcation are used to extract and match feature point pairs in two images with different perspectives.The experimental results verify that the accuracy and matching of feature points extracted based on the improved SIFT operator are higher than those of the SIFT operator and Harris operator.The feature points extracted and matched by means of the improved SIFT operator are more conducive to the 3D reconstruction of target objects.

关 键 词：物体特征提取和识别 SIFT RANSAC 三维重建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]