中文摘要：

在图像镶嵌算法中,SIFT算法本身就具有非常强的匹配能力,它可以处理两幅图像之间存在着的平移、旋转、尺度缩放等仿射变换,因此它是现今图像镶嵌中比较常用的方法。但在图像镶嵌中,如果直接使用SIFT算法会使运算量非常庞大,影响速度。针对这一问题,文中首先利用相位相关法来确定图像之间大致的重叠区域,再在大致的重叠区域提取SIFT特征点；其次利用特征点实现配准,并且在全局参数优化时考虑了径向畸变；最后用金字塔融合算法对图像进行融合。实验结果表明该算法能有效提高图像镶嵌的速度和精度,速度约提升三分之一,实现了一幅无缝的全景图像。

英文摘要：

In the image mosaic algorithm,SIFT matching algorithm has strong ability to handle the presence of translation,rotation,scaling affine transformation between two images,so it is now more commonly used image mosaic approach,but in the image mosaic,if using di-rectly SIFT algorithm makes computation very large,impacting the speed. Aiming at this problem,first determine the approximate region of overlap between the images by phase correlation method,and then extract the SIFT feature points in the overlapping area roughly,fol-lowed by the use of feature points to achieve alignment,and consider the radial distortion when optimizing the global parameters,finally pyramid fusion algorithm is used for image fusion. The experimental results show that the algorithm can be very effective in improving the speed and accuracy of image mosaic,the speed increased to about third,to achieve a seamless panoramic image.