摘要
提出了一种结合图像融合与小波变换的图像插值方法。首先利用一种常用图像插值方法获得初步放大图像,再将初步放大图像进行小波分解,源图像与小波低频子图进行直方图匹配后作为新的低频子图。将初步放大图像小波分解得到的不同方向的高频系数子图结合新的低频子图作为融合图像的多分辨率金字塔。最后进行小波逆变换得到融合图像,也即放大图像。实验结果表明该方法可提高图像插值的主客观效果,峰值信噪比相对常用方法可提高2dB以上。
An image interpolation algorithm combined image fusion and wavelet transform is presented. Firstly, an initial magnified image is obtained by one of conventional image interpolation methods. Then the initial resized image is decomposed with wavelet transform. The new low frequency submap is acquired by histogram matching original image and the old low frequency submap of the ini-tial resized image. The fused multiresolution pyramid is composed of high frequency submaps of the initial magnified image and the...
出处
《微计算机信息》
北大核心
2008年第9期312-314,共3页
Control & Automation
关键词
图像插值
图像融合
空间分辨率
小波变换
image interpolation
image fusion
spatial resolution
wavelet transform
