光学 精密工程, 2013, 21 (12): 3272, 网络出版: 2014-01-02
基于广义线性运算和双边滤波的红外图像增强
Infrared image enhancement based on generalized linear operation and bilateral filter
红外图像 图像增强 广义线性运算 双边滤波 infrared image image enhancement generalized linear operation Bilateral filter
摘要
针对传统的基于反锐化掩模法的红外图像增强方法噪声干扰较高且有光晕现象,提出了基于广义线性运算和双边滤波(BF)的红外图像增强方法。首先,通过双边滤波得到图像的基础层部分。接着,设计了广义线性运算模型,并在这个运算模型下对图像细节进行了非线性分割、去噪和自适应放大。最后,将经过动态范围压缩的图像基础层和自适应增强的图像细节层非线性叠加,得到最后的增强图像。另外,针对传统评价增强图像质量的平均对比度指标的缺点,提出了由局部到整体的改进平均对比度评估方法。3组对比实验观察和定量分析表明,本文提出的方法在对红外图像有效动态范围压缩和细节放大的同时,很好地抑制了红外图像的干扰噪声和光晕现象,得到的结果非常适用于实际红外热像仪的后端图像处理。
Abstract
As classical infrared image enhancement methods based on Unsharp Masking(UM) suffer from terrible noise interference and halo effect, a new approach based on generalized linear operation and Bilateral Filtering (BF) was proposed here. Firstly, the elemental part of the image was extracted by using the BF and then operation models based on generalized linear operation were designed for the following enhancement. In the new operation scheme, the image details were nonlinearly segmented, denoised and amplified and the dynamic range of the elemental part was compressed while the detailed part was enhanced adaptively. Finally, the enhanced image was obtained by adding the processed elemental and detailed parts together nonlinearly. Furthermore, an assessment method for image quality was presented. Unlike the classical metric which only considers the average contrast enhancement, the modified one involves both local and general average contrast enhancements. Both the comparison experiments and measurement index indicate that the proposed method can compress the dynamic range of infrared images, amplify their details, and suppress the noise interference and halo effect. This technique has a significant contribution to the posterior image processing of thermal infrared cameras.
贾宏光, 吴泽鹏, 朱明超, 宣明, 刘慧. 基于广义线性运算和双边滤波的红外图像增强[J]. 光学 精密工程, 2013, 21(12): 3272. JIA Hong-guang, WU Ze-peng, ZHU Ming-chao, XUAN Ming, LIU Hui. Infrared image enhancement based on generalized linear operation and bilateral filter[J]. Optics and Precision Engineering, 2013, 21(12): 3272.