摘要： 针对航空遥感成像范围小、视差角大的特点，提出一种航空遥感图像拼接算法。通过估算图像相对方位矩阵，使用透视变换实现图像校正，解决大视差角下的畸变问题。利用SIFT算法和基于概率密度的错误匹配点剔除方法，实现高精度图像配准，并通过小波变换完成图像融合工作。实验结果表明，在不同地表情况下，该算法的拼接性能均优于传统SIFT算法。

关键词: 图像拼接, 航空遥感, 透视变换, 小波变换, 精度评价, 图像校正

Abstract: Small imaging coverage and big parallax angle make aerial remote sensing more difficult to image processing. This paper proposes an algorithm for aerial remote sensing image mosaic. The original image is corrected by perspective transform with estimating matrix which is determined by the relative attitude of two images. It solves the distortion problem caused by big parallax angle. Image registration is approached by the Scale Invariant Feature Transform(SIFT) algorithm and a method based on probability density. The image mosaic process is completed by the wavelet transform. Experimental result shows that the mosaic performance of this algorithm is better than traditional SIFT algorithm in different surface conditions.

Key words: image mosaic, aerial remote sensing, perspective transform, wavelet transform, accuracy evaluation, image correction

中图分类号: 

• TP911.73