摘要
针对现有基于自然特征的增强现实系统中图像匹配准确度低、计算量大和鲁棒性差的问题,提出了一种基于仿射不变闭合区域和SURF(speeded-up robust features)的图像匹配方法。对输入的图像首先利用灰度直方图均衡进行图像增强得到二值化的图像,提取图像中的闭合区域作为图像的仿射不变区域,然后运用SURF检测算法提取闭合区域的图像特征描述,最后使用SURF双向匹配算法实现图像的匹配。实验结果表明,图像匹配的准确度有很大程度的提升,同时计算耗时更少;提出的方法能够满足增强现实系统的要求。
To solve the problem of low image matching accuracy, large amount of calculation and low robustness of the object on existing augmented reality system based on natural features, this paper proposed a new image matching algorithm based on affine-invariant closed region and SURF. First, the input image with histogram equalization for image enhancement to obtain a binary image and extracted the closed region in the image as the image of the affine invariant region. Second, it used SURF detection algorithm to extract the image of the closed region feature description. Finally, it finished the procedure of image matching using bidirectional matching with SURF algorithm. Experiment results indicate that the image matching accuracy largely improved while calculations consumed less time. The proposed method can meet the requirements of augmented reality systems.
出处
《计算机应用研究》
CSCD
北大核心
2014年第1期295-298,共4页
Application Research of Computers
基金
国家科技支撑计划课题(2012BAD32B04)
高等学校博士学科点专项科研基金资助项目(20120062110012)
关键词
图像匹配
特征匹配
闭合区域
仿射不变性
双向匹配
image matching feature matching closed area affine invariant bidirectional matching
