摘要
针对加速的具有鲁棒性特征算法导致匹配图像丢失颜色信息成分的问题,该文从图像匹配的效率和精度出发,提出了一种基于改进加速的具有鲁棒性特征算法和狄洛尼三角网的图像匹配方法。该方法在原有加速的具有鲁棒性特征算法的基础上,引入颜色不变量模型和狄洛尼三角网、三角形相似函数及摄影不变量等约束条件,有效保留了图像的颜色信息,减少特征点错误的匹配率。实验表明:本算法具有匹配速率高、提取的特征点多且分布均匀以及匹配率高等特点。
Image matching is one of the key technologies in the image processing field.From the point of view of image matching efficiency and precision,this article proposed a method of image matching based on SURF and Delaunay triangulation improvement combination.This method introduced the color invariant model and the constraint conditions including Delaunay triangulation,triangle similarity function and photography invariant into the original SURF algorithm.It effectively retained the image color information and reduced the erroneous matching rate of features.The experiment result showed that this improved method could have not only a high matching speed,but also many feature points which are well-distributed with good extraction performance.
出处
《测绘科学》
CSCD
北大核心
2015年第2期155-159,共5页
Science of Surveying and Mapping
基金
国家自然科学基金项目(41161069)
江西省教育厅科技项目(GJJ12384)
广西空间信息与测绘重点实验室开放基金(13-051-14-18)
