摘要
根据二元树复小波理论和图像插值的特点,将二元树复小波变换与基于边缘插值方法相结合,得到一个放大的插值图像,然后将插值后的图像进行一级小波分解,将分解后的高频子带再做小波变换,并修正变换后的高频子带系数,进行图像重构后得到最终的插值图像。实验结果表明,该方法能够提高图像的分辨率,同时消除边缘处的"振铃"效应。
Based on dual-tree complex wavelet theory and the characteristic of interpolation, this paper introduces a method, which combines dual-tree complex wavelet transform with new edge-directed interpolation. Firstly, we obtain an enlarged image by method of dual-tree complex wavelet transform and new edge-directed interpolation. Then, use wavelet to decompose the enlarged image, we take high frequency subband to wavelet transform, and modify high frequency coefficients of haar transformed. Lastly, wavelet reconstruction. The results show that the method can be used effectively to improve resolution of the images, as the same time elimination of images of the ＂ring＂ effect.
基金
内蒙古自治区自然科学基金项目(NO.200711020801)
内蒙古工业大学重点研究项目(NO.ZD200601)
关键词
二元树复小波
边缘插值
超分辨率
小波重构
dual-tree complex wavelet
edge-directed interpolation
super-resolution
wavelet reconstruction
