In this paper, an image processing method based on wavelet decomposition and image enhancement is proposed to hand of the issues that there is much noise and uneven distribution of light source and light intensity, in blast furnace infrared images. Firstly, Mallat algorithm is employed to decompose blast furnace infrared images into wavelet details and lower frequency part. Then, improved threshold filtering algorithm is presented to process wavelet details, and Retinex algorithm is used to adjust the intensity of the lower frequency part. Finally, high quality blast furnace infrared images with clear texture feature are obtained by imaging synthesis process of wavelet details and lower frequency part. As a result, the presented method is applied to the process of blast furnace infrared images in some iron enterprise and the efficiency and merits are shown.