基于引导滤波图像分层的红外烟尘图像增强

Infrared smoke image enhancement based on guided filtering image stratification

为解决烟尘环境对红外图像增强处理的干扰,突出目标的轮廓细节,提出一种基于引导滤波图像分层的红外烟尘图像增强方法。首先利用引导滤波将图像拆分为基础层与细节层,对细节层使用分数阶微分掩模作增强处理;然后基于红外烟尘图像的特点设计了二次分层方法,利用各项异性扩散将基础层分为原始层与轮廓层;之后对原始层进行自适应直方图均衡化,对轮廓层进行增益放大并与细节层合并;最后利用平均亮度设置权值函数,将两层图像进行加权融合得到增强图像。实验结果表明,相较于其他增强算法,该方法能够更有效的提高烟尘干扰下红外图像的清晰度,突出其细节纹理特征,增强后3组图像的平均梯度和信息熵平均值为7.7211及5.8114,相较于原始图像提升1.0119及3.1778。

To solve the interference of smoke environment on infrared image enhancement processing and highlight the contour details of the target,an infrared smoke image enhancement method based on guided filter image layering is proposed.Firstly,the image is divided into base layer and detail layer by guided filtering,and the detail layer is enhanced by fractional differential mask.Then,based on the characteristics of infrared smoke image,a secondary stratification method is designed.The base layer is divided into original layer and contour layer by anisotropic diffusion.Then adaptive histogram equalization is performed on the original layer,and the contour layer is amplified and merged with the detail layer.Finally,the average brightness is used to set the weight function,and the two layers of images are weighted to obtain the enhanced image.The experimental results show that compared with other enhancement algorithms,the proposed method can more effectively improve the clarity of infrared images under smoke and dust interference and highlight their detailed texture features.The average gradient and information entropy of the three groups of images after enhancement are 7.7211 and 5.8114,which are 1.0119 and 3.1778 higher than the original images.