本文提出了一种基于特征点的全自动无缝图像拼接方法。该方法采用对于尺度具有鲁棒性的SIFT 算法进行特征点的提取与匹配，并通过引导互匹配及投票过滤的方法提高特征点的匹配精确度，使用稳健的RANSAC 算法求出图像间变换矩阵H 的初值并使用LM 非线性迭代算法精炼H，最终使用加权平滑算法完成了图像的无缝拼接。整个处理过程完全自动地实现了对一组图像的无缝拼接，克服了传统图像拼接方法在尺度和光照变化条件下的局限性。实验结果验证了方法的有效性。-This paper presents a feature points based on automatic seamless image mosaic method. The method used for the scale of the SIFT algorithm is robust feature points extraction and matching, and guide each other through the match and voting filtering feature point matching method to improve the accuracy of the RANSAC algorithm for the robust transformation matrix H between the images obtained The initial and refined using the LM non-linear iterative algorithm H, the final completion of the weighted smoothing algorithm for seamless image. Completely automated the entire process of a group to achieve seamless image stitching, image stitching method to overcome the traditional scale and illumination changes in the conditions of the limitations. Experimental results demonstrate the effectiveness of the method.