周克虎, 雷涛, 罗刚. 一种基于时域滤波的红外序列图像去噪算法[J]. 应用光学, 2021, 42(3): 474-480. DOI: 10.5768/JAO202142.0302004
引用本文: 周克虎, 雷涛, 罗刚. 一种基于时域滤波的红外序列图像去噪算法[J]. 应用光学, 2021, 42(3): 474-480. DOI: 10.5768/JAO202142.0302004
ZHOU Kehu, LEI Tao, LUO Gang. Infrared sequence images denoising algorithm based on temporal filtering[J]. Journal of Applied Optics, 2021, 42(3): 474-480. DOI: 10.5768/JAO202142.0302004
Citation: ZHOU Kehu, LEI Tao, LUO Gang. Infrared sequence images denoising algorithm based on temporal filtering[J]. Journal of Applied Optics, 2021, 42(3): 474-480. DOI: 10.5768/JAO202142.0302004

一种基于时域滤波的红外序列图像去噪算法

Infrared sequence images denoising algorithm based on temporal filtering

  • 摘要: 红外图像是现代光学设备常用的图像源,图像显示效果直接影响设备的用户体验,而红外序列图像中的噪声会导致显示效果的下降。为了减轻噪声对红外序列图像显示效果的影响,通过历史多帧灰度值的加权和当前帧无噪声图像的估计,提出了一种基于时域高斯滤波的去噪方法。参考空域双边滤波的权值分配方法,引入了灰度值的影响对时域高斯滤波的权值进行修正,解决时域滤波导致的序列图像中运动目标拖尾和模糊。实验结果表明,时域滤波方法能够有效平滑帧间噪声,减轻噪声导致的红外序列图像显示效果的恶化,引入灰度值的影响进行滤波权值修正之后,能够解决时域滤波导致的运动目标拖尾和模糊问题。

     

    Abstract: The infrared image is a commonly used image source of modern optical equipment, the display effect of the image directly affects the user experience of the device, and the noise in the infrared sequence images can cause the decline of display effect. In order to reduce the effect of noise on the display effect of infrared sequence images, by weighting the gray values of multiple historical frames and estimating the noise-free image of the current frame, a denoising algorithm based on the temporal Gaussian filtering was proposed. Referring to the weight distribution method of spatial bilateral filtering, the influence of the gray value was introduced to correct the weight of temporal Gaussian filtering, which solved the trailing and blurring of moving targets in sequence images caused by the temporal filtering. The experimental results show that the temporal filtering method can effectively smooth the noise between frames and reduce the deterioration of display effect of the infrared sequence images caused by the noise; and the problem of trailing and blurring caused by the temporal filtering can be solved with the influence of gray value which is introduced to correct the filtering weight.

     

/

返回文章
返回