基于奇异值分解和非下采样Contourlet变换的红外图像增强新算法

A New Infrared Image Enhancement Algorithm Based on Singular Value Decomposition and Nonsubsampled Contourlet Transform

针对夜间红外图像噪声大、对比度较低的问题,本文提出一种基于奇异值分解和非下采样Contourlet变换的夜间红外图像增强新算法。通过分解高频系数奇异值,对得到的奇异值矩阵进行权重排序分析,保留大权值的高频系数进行降噪;再使用自适应权值的增益函数对降噪处理后的高频系数进行增强处理。对低频系数则使用自适应权值的改进Sigmoid函数进行处理,不仅可提升图像整体对比度信息还可减少传统Sigmoid函数存在的过增强问题。实验结果表明,该算法能有效抑制红外图像噪声,提高图像对比度信息。

Considering the problems of infrared image, an infrared image enhancement method based on singular valuedecomposition and NSCT transform was proposed. In order to de-noise the high frequency coefficient by using the weightsorting method of singular value, the singular value of high frequency coefficients was decomposed, and the weight of thesingular value matrix was sorted to retain the high frequency coefficient of large weights. Then, the gain function of adaptiveweights was used to enhance the high frequency coefficient after noise reduction. In addition, the improved Sigmoid functionof adaptive weights was used to deal with the low frequency coefficients, which could increase the image contrastinformation and reduce the over-enhancement of the traditional Sigmoid function. Experiments showed that the algorithmcould effectively suppress infrared image noise and improve the image contrast information.