论文部分内容阅读
无人机在飞行过程中由于机体的倾斜和抖动,造成航拍图像出现大的仿射变形。因此,传统的图像拼接算法很难得到好的效果。基于SIFT(Scale Invariant Feature Transform)特征的图像拼接算法,首先通过提取图像的尺度不变特征点,解决了待拼接图像间大的平移、旋转、尺度变化的干扰。然后,通过欧式距离判断得到初匹配特征点集,并利用RANSAC方法进一步精确了匹配点集,得到了准确的变换矩阵;最后,采用加权平均法的图像融合技术,得到了稳定的、鲁棒的图像拼接结果。
UAV in flight due to body tilt and jitter, resulting in large affine deformation of aerial images. Therefore, the traditional image mosaic algorithm is difficult to get good results. Based on SIFT (Scale Invariant Feature Transform) feature, the image splicing algorithm firstly solves the problem of large translation, rotation and scale variation between images to be spliced by extracting scale-invariant feature points of the image. Then, the initial matching feature point set is obtained through the judgment of the Euclidean distance, and the matching point set is further refined by using the RANSAC method to obtain the accurate transformation matrix. Finally, the image fusion technique of the weighted average method is adopted to obtain a stable and robust Image mosaic results.