Wavelet transform method has been widely used in image filtering, the wavelet threshold de-noising method can treat Gaussian noise with randomness well. This paper proposes that after the wavelet transform the high frequency coefficients need a more accurate processing, And the classical hard threshold method has been improved by introducing the measure of medium truth scale. The new method can effectively handle strong Gaussian noise with larger variance through theoretical analysis and experimental simulation, and get a fine recovery image. It also provides a new approach for wavelet de-noising.
展开▼