摘要
传统的内插算法在放大红外图像时都存在着一定的缺陷,提出了一种基于小波重构和灰度分段变换的图像放大新算法。该算法先对原始图像进行小波变换获得高频系数,运用牛顿插值算法放大高频系数作为放大图像的高频成份,再将原始图像作为低频成份,进行小波重构,可得放大图像。为了增强放大图像,将图像按双灰度阈值分割成对应目标的灰度值高段、对应背景的灰度值低段和对应过渡区域的灰度值中段等3个部分,对各部分采用不同的线性变换,获得最佳的视觉效果。实验证明该方法在图像细节方面具有很好的放大效果。
For an infrared image, traditional interpolation algorithms have their faults. A new algorithm for infrared image magnification and enhancement based on wavelet transform and gray-level section transforms is proposed. The high-frequency coefficients obtained from the initial image by wavelet transform are enlarged through Newton interpolation, which are used to be high-frequency components of the magnification image. The initial image is used to be the low-frequency component of the magnification image. Then the magnification image is reconstructed by wavelet inverse transform, in order to enhance the magnified infrared image. The image gray levels are divided into three sections by bi-threshold that is average value of image gray level. Three sections are that high one corresponding to target, low one to background, and middle one to transitional region. For the best result of vision, different transforms are taken for different sections. The test demonstrates that the algorithm has a very good magnification effect in image resolution detail.
出处
《红外技术》
CSCD
北大核心
2008年第10期567-570,共4页
Infrared Technology
基金
国家自然基金资助项目(编号:60377034)
关键词
图像放大
小波变换
图像增强
灰度分段
image magnification
wavelet transform
image enhancement
gray-level section
