摘要
针对传统的小波和偏微分方程(PDE)算法无法准确估计图像的平坦区域和边缘区域而出现虚假边缘,以及小波低频估计不准确导致的图像模糊,提出一种小波与改进的PDE插值结合的图像超分辨率重构算法。该算法首先对传统的PDE算法进行改进,提出一种加权拟合的PDE插值法对图像边缘实现较好的定位;然后利用小波提取高频成分并调整其系数,并将增强幅度后的原始图像作为低频部分,进行小波逆变换获得高质量的重建图像。实验结果表明:这种方法可以充分发挥两种算法的优点,不仅在提高图像分辨率的同时较好地保留了原始图像的细节信息,还提高了放大后图像的亮度和清晰度。
In view of the traditional wavelet and partial differential equation （ PDE ） algorithm can not accurately estimate the image smooth, edge area or low-frequency, which caused false edge and image blurring, a wavelet and improved PDE interpolation of image super-resolution reconstruction algorithm was proposed. Firstly, the traditional PDE was improved by a weighted PDE algorithm to achieve better positioning for image edge. Then, High-frequency parts were extracted by wavelet transformation and their coefficient were adjusted. The enhanced amplitude of original image was taken as the low-frequency parts. Finally, a high resolution image was achieved by wavelet transformation. The experimental results showed that this method had the advantages of the two algorithms, not only improved picture resolution and effectively kept the detail information of the original image, but also improved the definition and brightness of zoomed image.
出处
《探测与控制学报》
CSCD
北大核心
2011年第5期72-76,共5页
Journal of Detection & Control
关键词
图像重构
超分辨率
小波变换
PDE插值
数据融合
image reconstruction
super-resolution
wavelet transform
PDE interpolation
data fusion
