小波域内循环平移操作的图像高分辨率重建算法

为解决传统的图像插值算法因具有全局性而不能较好地处理图像边缘细节信息，且易在细节区域产生锯齿线的问题，提出了一种图像分辨率和对比度增强算法。该算法先用小波零填充算法得到高分辨率图像，并通过纠正残差过程来弥补丢失的边缘和纹理特征，然后对其进行定向循环平移操作。考虑到图像小波分解后水平、垂直、对角方向的高频分量能够反映图像这3个方向的边缘变化情况，从而利用图像不同方向的高频分量来刻画图像像素点不同方向的突变程度。根据这个突变程度来实现循环平移操作的自适应融合过程，这样可以避免过度抑制边缘细节信息。最后对重建的高分辨率图像小波分解后的高频分量使用非线性增强函数，提高图像对比度，突出边缘和轮廓信息。实验结果表明，该算法在增强图像空间分辨率和对比度的同时，保留了原图像包含的边缘和轮廓信息，不仅有较好的视觉效果，还有一定的抗噪能力。

Image High Resolution Reconstruction Algorithm Using Directional Cycle Spinning Operations in Wavelet Domain

Traditional image interpolation algorithm is good at global handling of images, but weak in dealing with edge details of images, thus leading to sawtooth lines in detail areas. For this problem, this paper proposes an image resolution and contrast enhancement algorithm. First, a high-resolution image is obtained from an original one by using wavelet zero padding algorithm, and the lost edge and texture features of the image are compensated by using residual error correction process. Then, directional cycle spinning operations are performed on the image. Considering that the high frequency components in horizontal, vertical and diagonal directions of the image after wavelet decomposition can reflect the edge changes in the image, we use high frequency components in different directions of the image to describe mutation degrees in corresponding directions of image pixels. According to the mutation degrees, an adaptive fusion process of cycle spinning operations is realized, which can avoid excessive suppression of edge details. Finally, a nonlinear enhancement function is used to improve image contrast and highlight edge and contour information, Experimental results show that this algorithm not only enhances the spatial resolution and contrast of images, but also retains the edge and contour information contained in original ones. Compared with other image interpolation algorithms, this algorithm shows improved performs both in visual effect and anti-noise ability.