一种基于SIFT特征匹配的自动图像拼接方法

An Automatic Image Mosaic Method Based on the SIFT Feature Matching

提出了一种利用尺度不变特征变换匹配技术应用于图像自动拼接的方法，SIFT特征是图像的局部特征，其对旋转、尺度缩放、亮度变化均保持不变，对视角变换、仿射变换、噪声均有一定程度的稳定性。在分析了SIFT特征匹配理论的基础上，利用SIFT获取匹配点对，结合最小二乘匹配方法提高了匹配的精度，通过计算图像之间的变换参数，实现图像的自动拼接。实验取得了较好的结果，验证了方法的可行性。

This paper provides a SIFT feature matching method to the image mosaic. The SIFF feature is invariant with the rotation, scale zoom, gray change, and is stable for the visual angle change, affine transformation and the noise. The paper analyzes the detailed academic theory of the SIFT, after obtaining the matching points with the SIPT and improving the matching accuracy with the LSM method, the image mosaic is accomplished with the distortion parameters. Experiments prove the effectiveness of the proposed image mosaic method.