Abstract: In view of the characteristics of radiographic image, including high noise, poor contrast, image blur, and so on, an adaptive radiographic image denoising algorithm using multiscale products threshold is proposed in this paper. It can overcome the conventional radiographic image denoising algorithms' problems of poor denoising effect, blurring image and losing defect edges and details. In this algorithm, the ideas of noise level estimation, multiscale, products threshold and wavelet transform are skillfully used, and then the denoised image with high quality is obtained. Taking the real radiographic images of industrial weld testing for example, the denoising comparison experiments are performed between the proposed algorithm and the conventional algorithms, including wavelet denoising, median filtering, Wiener filtering and wavelet-median filtering. Moreover, the experimental results demonstrate that the proposed algorithm not only has excellent denoising performance, but also preserves the defect edges and details well.