Abstract: The medical CT Image will bring in noise when in the process of imaging. The noise can reduce the quality of image and affect the clinical diagnosis. Therefore, medical CT image denoising plays an important role in the diagnosis and treatment. This paper proposes the classical NL-Means algorithm at firse, then changes the denoising key role of the weighted kernel function. The experiment shows the algorithm for medical CT image denoising have certain effect, and can better preserve the detail information. For the PSNR, it can be improved by 2.59 dB at most.