针对工业检测中用传统的降噪方法难于进行实时处理的问题,提出了一种基于图形处理器(GPU)加速的动态降 噪方法。借助于显卡中图形处理器较强的运算能力，利用CUDA并行计算架构在PC机上实现了图像的动态降噪。此方 法用软件实现，可大大提高处理速度，降低成本，易于修改。实验结果表明,该方法具有较好的实时性和去噪效果。

In accordance to the fact that real time processing is difficult to be implemented by using traditional noise reduction methods in industrial inspection, a dynamic noise reduction method based on Graphic Processing Unit (GPU) acceleration for infrared images is proposed. With the help of the strong computing ability of the GPU in a display card and the parallel computing architecture of a CUDA, the dynamic noise reduction for infrared images is implemented on a personal computer. This method is realized by using a software. It has the features of high processing speed, low cost and modifiableness. The experimental result shows that the method is excellent in real time processing ability and noise reduction.