一种有效的全景图像自动拼接算法

An effective automatic mosaicing algorithm of panoramic images

图像拼接是基于图像绘制技术中的一种常用方法。该文用特征提取算子得到2幅图像的特征点,根据特征点邻域灰度的归一化相关性得到初始的匹配点集,利用特征点与其周边特征点之间的关系来获得更加精确匹配;利用随机采样一致性方法对匹配点集进行优化,以具有最多内点数的模型作为投影矩阵的系数。

In this paper, an effective image mosaic method is presented. The feature detectors are used to obtain the feature points. The initial match points are found by normalized cross correlation. Then the matches are refined according to the relationship of neighboring feature points. Random sample consensus is introduced to optimize the matches, and the model which has the most inpoints is used as the transform coefficient