基于NSCT的图像拼接算法

Image stitching algorithm based on NSCT

为了提高图像拼接的速度和拼接图像的视觉效果,提出了一种基于非下采样轮廓波变换的图像拼接方法。首先利用角点检测算法获得待拼接图像的匹配点,选取以匹配点为中心的空间频率最大的一定大小的子图像为模板,利用流形上的优化算法求解变换参数;再将待拼接图像分解为不同尺度和不同方向的子频带图像,在变换域进行对应的拼接;最后利用重构算法将拼接的各个子频带图像进行反变换,得到最终的拼接图像。实验表明,此方法速度更快,视觉效果更好,拼接后的图像更加自然,过渡更平滑。

In order to improve the visual effect of stitching image and stitching rate,a new stitching method based on Nonsubsampled Contourlet Transform （NSCT） was proposed.Firstly,corner detecting method was used to obtain the matching points and the sub image whose space frequency was maximum was selected as the model.Then the source images were decomposed into sub band images of different scales and different directions using NSCT.Stitching the sub band images in the transform domain,then inverse transform of NSCT is performed to reconstruct the final image from different sub bands.The experimental results show that the new method can achieve superior performance in stitching rate and visual effect.The result image is more natural and the transition is much smoother.