基于时域递归滤波的动态数字图像降噪

Noise-reduction of Digital Video Image Based on Time-recursive Filtering

根据时域递归滤波的原理,得出递归滤波的本质就是加权平均的低通滤波器,以及滤波系数与信噪比改善和拖影时间之间的关系曲线。分析了递归滤波算法用于降低数字动态图像随机噪声的优缺点,并结合自适应控制理论,提出解决滤波中拖影问题的方法。在此基础上,设计了相应的基于FPGA的硬件及软件,取得了预期的试验效果。

With the analysis of the principle of time-recursive filter, the authors find that the nature of time-recursive filter is Weighted Mean Low-Pass filter and the relationship between filter coefficients and SNR and the artifact arising from the area where motion can not be tracked. One approach based on Automatic Control Method is proposed to solve the artifacts problem during the process of filtering and the authors designed the corresponding hardware based on FPGA and software respectively. The experimental results show that the time-recursive filtering is good at digital video image denoising and reduces the random noise of digital camera system.