基于Levenberg-Marquardt算法的图像拼图研究

Research on Image Registration Based on Levenberg-Marquardt Algorithm

提出一种基于LM(Levenberg-Marquardt)算法的图像拼图方法。图像拼图是将不同时间段、不同视点获取的2幅或多幅图像拼接在一起。算法首先利用手动选取的对应点,基于线性最小二乘算法,估计出射影变换矩阵,然后利用LM优化算法,基于平方误差累加和目标函数进行最小化,当获得最优射影矩阵,利用最采样技术,将两幅图像拼图。基于真实图像实验,说明算法是简单可行的。

This paper proposes an image registration method based on Levenberg-Marquardt algorithm. Image registration is the process of overlaying two or more images of the same scene taken at different times, from different viewpoints. Based on linear least squares regression, the method first estimate a initial projective transformation matrix with the points given by the user, and then Levenberg-Marquardt optimization algorithm is applied to minimize the sum of squared errors. Once the algorithm finds the best projective transformation matrix, two image be warped to generate image mosaic using image resampling method. The effectiveness of the method is demonstrated by experiments on real images.