Research On Insulator Infrared Image Denoising In The Wavelet-domain Based On Gaussian Mixture Model

Infrared technology has been applied widely to monitor the high voltage insulator in electric power system. However, the insulator infrared image is always contaminated by noise. Therefore how to eliminate noise effectively has become an important pretreatment step.The key point of the image denoising algorithm in the wavelet domain is how to select a suitable statistical model of wavelet coefficient. The study shows that the coefficients of image wavelet transform subject to the Alpha stable distribution. Unfortunately, the closed-form expression for the Probability Density Function (PDF) of the general SaS signal does not exist, and the characteristic function for the probability distribution of the SaS signal is discontinuous at some points. These will bring inconvenience to the signal analysis and the application. Based on the theory of the Alpha stable distribution, In this paper, the high frequency coefficients of the infrared image were modeled with Gaussian mixture distributions (GMM) which is an approximate expression of Symmetric Alpha Stable (SaS).In addition, the maximum a posteriori (MAP) estimator was derived. The experiment result shows that the quality of the image improves greatly, and the proposed method is much less complex.