Tire images are important data for realizing intelligent tire manufacturing tasks. However, tire images usually suffer from noise in the process of shooting and other collection stages. The noise results in the loss or distortion of detailed information in the image, thereby the accuracy and reliability of tire image are reduced. For decreasing the noise in tire images, the paper presents an effective tire image denoising method based on total variation algorithm. First, the tire noise images are constructed. Salt & pepper noise, Gaussian noise, gamma noise and Rayleigh noise models are utilized to establish tire noise images. Then, Total Variation (TV) algorithm is applied to implement the tire image denoising. It can remove noise, while the tire image’s contours and details are preserved. Finally, the SNR values of five types of denoising algorithms are calculated and compared for analyzing the denoising performance. Experimental results show that our method can reach optimal or suboptimal SNR values for each noise models.