Study on robust noise reduction algorithm based on wavelet transfrom

Based on the principle of wavelet transform, the basic assumptions of wavelet denoising is analyzed. In this paper we propose a robust wavelet transform algorithm using a α - trimmed mean filter as a “agent” component based on Donoho´s soft thresholding shrinkage method. And on this basis, a detailed analysis on the structure of two-dimensional wavelet transform is made; two-dimensional robust wavelet transform algorithm is built, and applied to image noise reduction. One-dimensional signal with noise and gross error test results show that the effectiveness of robust wavelet transform. This paper also uses image analysis with salt and pepper noise, compared to traditional wavelet denoising algorithm. The results show that the algorithm can effectively remove impulse noise in the image, but it will make the image fuzzy if the gross error threshold is set to some extent inappropriate.