Abstract: Based on the red-black wavelet transform,a new thresholding for image denoising is proposed.Combing the characteristics of affine and non-negative thresholding,this method can keep a better balance between image details and it's denoising.After simulation on a multitude of images,eight different thresholding weights applied to as many as eight levels of red-black wavetlet transforms have been proposed.Simulaton proves that by combining the thresholding and the eight weights better results can be achieved.