Research On Image Denoising With Wavelet Transform

As an important component of signal processing and modern communications, the technology of image denoising is getting more and more close with people's relati- ons. In the applications, as the processing and recognition of the image it is often the foudation of the analysis and the processing of the image in a higher level. Wavelet transform is a time-frequency analysis method which develops on the basis of Fourier transform,and it is an effective analytical tool. It has the low entropy,the multi-resolution, decorrelation, the flexible election-based and so on features and the superiotity which has in the signals aspect, the above enables it to obtain the widespr- ead application in the imagery processing domain. Among them, the processing of no- isy image using wavelet transform can effectively filter out the noise, keep signal's high-frequency information, and make the best of recovery of the original signal.The traditional wavelet threshold denoising method can eliminate the noise to some extent, but the image has fuzziness inordinately. This paper proposed two wave- let denoising methods. One is the denoising method based on the mean filtering and the wavelet transformation, the two combinative methods can fully improve the filter performance better using the respective merit binding energies use the fully, and can also eliminate the image effectively while maintaining the edge information of the image.This method is an effective image denoising method. Another was aimed at traditional the threshold value function existence insufficiency to make the improvem- ent to the traditional wavelet threshold function, simultaneously also proposed the new wavelet contraction threshold value based on the image strange characteristic.Comparing with several other methods, the matlab simulation experiment has confirmed that the two improved wavelet methods for the noise removal in this article have advantages in the feasibility, the validity and the superiority.