Taking the infrared images of water jet as the research object, a multistep denoising scheme for infrared images is proposed. Firstly, an image subtraction method is used to obtain the increment distribution of the infrared temperature field. Secondly, an adaptive median filter is used to remove the white noise with Gaussian distribution. Finally, a Gaussian high-pass filter is used to remove the periodical noise and sharpen the image. The experimental result shows that the proposed multistep denoising method is good in denoising. In the processed image, the structural characteristics of the jet core, main region, diffuse region and different turbulent jet in the turbulent flow composed of water mist and flowing air induced by the momentum exchange between the jet boundary and ambient air are revealed clearly. The analysis of the denoised infrared images helps us further understand the mechanism of turbulent jet. The scheme is of significance to the processing of similar infrared images, too.