基于改进SURF和RANSAC算法的室内环境图像匹配

Indoor Environment Image Matching Based on Improved SURF and RANSAC Algorithm

针对移动机器人在室内环境下定位及地图创建要求,比较SURF特征算法以及随机抽样一致性(RANSAC)算法各自优势,设计实现以改进SURF及RANSAC相结合的室内图像匹配方法.初始阶段,采用SURF算法提取图像特征,利用双向K最近邻分类法筛选提取到的特征点进行特征预匹配;再利用RANSAC算法对预匹配结果进行优化,剔除误匹配对,完成对图像的匹配矫正;最终得到准确的匹配图像.实验结果表明,此方法提高了匹配的正确率,也缩短了匹配时间,提高了匹配效率.

In view of the requirements of binocular vision robot localization and navigation in the complex indoor environment,and compared the advantages of Speeded-Up Robust Features and Random Sample Consensus,and a matching method of indoor environment image put forward based on SURF and RANSAC algorithm. Firstly,f eature of images deteced and pre-matched by using SURF. Secondly,using RANSAC algorithm to optimize the matching process,complete the correction of image matching and eliminate false matching. Last,get the images. The experiment verified the feasibility of the method,also has a good application effect in the indoor environment.