The application of an infrared temperature measurement system reduces the occurrence of safety accidents during manual temperature measurement. However, the accuracy of the measurement depends on the quality of the image obtained using the infrared thermal imaging camera. To reduce the influence of noise on the quality of molten steel infrared images, this paper proposes a denoising method based on adaptive Wiener filtering. The autocorrelation parameter exponential decay model is used to control the computational complexity and sensitivity of the algorithm, thereby effectively improving the denoising performance of the Wiener filter. Based on the denoising processing of molten steel infrared images at different temperatures, it is verified that the proposed denoising method has better denoising performance than Wiener filtering and sparse decomposition methods.