Abstract: A mosaic method for aerial video based on the block Harris feature was proposed, which is able to cope with the problem of high computation and low accuracy of the current mosaic methods. This method overcomes the shortages of the traditional SIFT method, since it can save the matching time and improve the accuracy. First, block Harris feature extraction method is used to obtain a better distributed of the feature point. Then, pyramidal Lucas-Kanade algorithm is used to match the feature points. After that, an improved RANSAC method is used to calculate the affine parameter accurately. Experimental results show that our method can do mosaic quickly and correctly.

Key words: aerial video, image mosaic, SIFT feature, Harris feature, RANSAC algorithm

摘要： 针对现有航拍视频拼接算法处理速度慢、准确性不高等问题, 提出一种基于分块Harris特征的航拍视频拼接方法, 改进了传统基于SIFT特征提取算法的不足, 缩短了匹配时间, 提高了匹配准确性。首先采用分块Harris角点提取的方法均匀提取图像中的角点, 然后采用金字塔光流算法进行角点匹配, 最后通过改进的RANSAC方法求出仿射变换参数。实验表明, 该方法能够实时对航拍视频进行拼接, 具有更高的准确性。

关键词: 航拍视频, 图像拼接, SIFT特征, Harris特征, RANSAC算法