摘要
提出了一种基于全变差和对比度的红外图像变分增强方法.针对红外图像噪声较大的特点,在现有变分增强算法的基础上,构造目标梯度场时,事先考虑了对噪声的限幅,防止不必要的噪声放大.另外,构造目标函数时加入了图像的全变差(TV)约束,使得噪声进一步得到抑制.直方图类算法会对灰度邻近、分布较少的灰度阶进行合并,从而丢失一些有用细节.本算法不会出现直方图算法的类似问题.实验表明本算法相对于常见的直方图处理算法,可以更有效增强红外图像.
A variational enhancement method for infrared images was proposed based on total variation （TV） and contrast. Considering that the infrared images generally are much noisy, the proposed method restricts the noise to compute the target gradient based on the existing variational enhancement scheme. In addition, the object function includes an extra term representing total variations, which gives a further suppression to the noise. The histogram-based algorithms are likely to merge the gray levels which are similar in intensity and have low distribution, thus some useful details will be decreased. The proposed variational method does not have such problems as histogram method, and is more effective to enhance infrared images.
出处
《红外与毫米波学报》
SCIE
EI
CAS
CSCD
北大核心
2006年第4期306-310,共5页
Journal of Infrared and Millimeter Waves
关键词
红外图像
图像增强
变分
全变差
直方图
infrared image
image enhancement
variational method
total variation （TV）
histogram
