Abstract: To identify the noise types of CT image,corresponding method was used to improve the effect of image denoising,reduce the blindness.By analysis the energy distribution of the HH subband's coefficients in the wavelet domain,SNR of the histogram and area of curve fitting was used to distinguish Gaussian white noise and salt & pepper noise in computed tomography images.We found histogram of the signal to noise ratio R equals 0.2,curve fitting map area A equals 60 are the distinguish lines between Gaussian white noise and salt & pepper noise.The experiments on a wide variety of images show the veracity of this method in distinguishing noise.