欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
Harris是一种高效的角点检测算法,但不具备尺度不变性。SURF（speeded-up robust features）算法虽然能很好地解决图像尺度变化问题,但是在特征点提取方面没有Harris稳定。针对Harris和SURF两种算法的特点,提出一种新的Harris-SURF特征点提取算法。首先用Harris算法检测图像角点,再用SURF算法提取图像特征点;然后合并角点和特征点,并剔除重复点获得新的特征点集,确定新特征点的主方向并生成特征描述符,再对图像使用比值法进行初匹配;最后利用RANSAC剔除错误匹配点实现精确匹配。实验结果表明,该算法对图像存在旋转、缩放、光照及噪声变化有较强的鲁棒性,同时提高了运行效率。
英文摘要：
Harris is an efficient corner detection algorithm,but it doesn＇t have the scale invariance. SURF algorithm can solve the problem of image scale changes,but it is less stable than Harris in respect to feature point extraction. This paper proposed a new Harris-SURF feature point extraction algorithm according to the characteristics of Harris and SURF algorithm.Firstly,it extracted image corners using Harris algorithm and detected image feature points using SURF algorithm,then it merged corner points and feature points and eliminated duplicate points to obtain a new feature point set,and determined main directions of feature points and generated feature descriptors,then used ratio method to get initial matching. Finally,it used RANSAC to eliminate errors and achieve accurate matching. Experiments show that the algorithm has strong robustness for image with rotation,scaling,illumination and noise changes and improve efficiency.
同期刊论文项目
同项目期刊论文
期刊信息
位置：