Abstract:In this paper new manner for removing noise from image using wavelet fusion method. The main aim of this work is restaurant the image based on peak signal to noise ratio measure. The key idea is compared each sub band for different levels of wavelet based on PSNR value. Initially apply discreet wavelet transform with 2level decomposition on the set of images .Then perform denoising wavelet techniques that achieved by threshold value for detail coefficient and compare it with wavelet coefficients for detail sub band. After that select sub band that has less noise from each image, sub band that contain high PSNR measure is the optimal. Finally apply IDWT process to convert the result image from frequency domain to spatial domain. The outcomes of the work exposed that the number of levels increases, PSNR of image decrease. In this paper was chosen two level of decomposition to guarantee choosing several sub band for fusion process, but the increasing in number of levels of wavelet will lose the essential information of image, therefore level 1 is better than level 2.

Conclusions
This paper presents removing noise from images using wavelet fusion method .Through the outcomes of this work we notice that PSNR value was decreased as number of levels was increased. This work reduces noise at different resolution levels. This work concludes PSNR value of level one is higher from PSNR of level two. Also the results show that soft thresholding give value large of PSNR than hard thresholding.