摘要
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。
During the flight of unmanned Aerial Vehicle (UAV), the aerial images taken by it can be greatly affined-deformed due to tilting and shaking of UAV. Therefore, it is very difficult to attain the good result by using conventional images mosaic algorithm. This paper details the images mosaic algorithm characterized with Scale Invariant Feature Transform (SIFT), which solves the problem of interference for great translational motion, rotation and scale variation between the images to be mosaicked by firstly extracting the scale invariant feature points of the images, then the initial matching feature points set is attained by means of Euclid's distance judgment, and further précised with RANSAC method so as to work out the accurate transformation matrix, finally the image blending technique in weighted average method is adopted to achieve the stable and robust image mosaic effect.
出处
《教练机》
2011年第4期9-14,共6页
Trainer
