一种基于相位相关优化的图像拼接方法

A Image Mosaic Method Based on PCT Combined with Optimization

针对图像拼接中的图像配准参数估计问题,提出了一种扩展相位相关法与优化相结合的参数估计方法。该方法首先采用扩展相位相关法估计出相互间存在旋转和平移的两幅部分重叠的图像的初始变换参数,然后利用LEVENBERG-MARQUARD优化算法对参数进行优化而得到最终准确值,最后根据这一参数值采用无缝拼接技术将两幅图像拼接起来,通过实验表明该方法是有效的且效果比较好。

By analyzing some disadvantages of initial parameter estimate in image registration ,We present a phase correlation technique combined with optimization image mosaic method. Here, let two images are different only by a displacement and a rotation, then we can measure the initial parameter estimate vector with the phase correlation technique. Subsequently by optimizing this initial parameter estimate vector with Levenberg-Marquard algorithm ,we can atlain finally accurate data. Finally, we may stitch the two image in accordance with this parameter. Experimental results have demonstrated this method to is effective.