Image Denoising Method Based on Improved Wavelet Threshold Transform

An improved denoising algorithm based on a new threshold adjusted by a coefficient considering noise variance, and a new threshold function is proposed against the problems existing in current wavelet transform, such as that the fixed threshold formula does not follow the changes of the decomposition scale, the hard threshold function is not continuous at the threshold point, the constant deviation between the original wavelet coefficients and their estimates always exists in soft threshold function, and so on. This paper firstly layered the noisy images with wavelet functions. Then, threshold is calculated with the new threshold formula and the noise is removed through the new threshold function. The threshold is dynamically adjusted in the time-frequency domain and the deviation between the practical and the estimated wavelet coefficients is decreased by adding an adjustable parameter in the new threshold function. Finally, the processed image is reconstructed. Simulation results show that the denoising effect of the proposed algorithm is better than that of traditional wavelet threshold denoising.