基于加速鲁棒特征的广角图像自动拼接校正算法

广角图像在全景图拼接中优势明显,然而,由于广角图像存在严重的透视畸变,往往导致拼接的效果不理想,甚至导致拼接失败。针对广角图像拼接中的透视畸变问题,提出一种广角图像自动拼接、校正算法。先计算出待拼接广角图像的特征点匹配对,求出图像拼接矩阵,对广角图像进行拼接;然后,根据图像采集时的镜头旋转角度计算出图像校正矩阵,对拼接后的全景图进行透视畸变校正。实验结果表明,基于该算法得到的全景图能够有效地消除透视畸变,达到较好的拼接效果。

Automatic stitching and regulating algorithm for wide-angle images based on speed-up robust feature

Wide-angle images are very helpful when stitching panoramas.However,the severe deform in wide-angle images usually results in stitching failure.Focusing on such a problem,an automatic stitching and regulating algorithm for wide-angle images was proposed.The matching pairs of feature points were calculated to obtain the stitching matrix at first in order to stitch wide-angle images.Then the correcting matrix for the image was calculated based on the rotating angle of the lens so that a regulation could be made on the perspective distortion.It is shown from experiments that the perspective distortion can be effectively removed from the panoramic view obtained with the presented algorithm.