In view of the drawbacks of traditional wavelet threshold method for image noise reduction that it cannot remove the noise thoroughly,and may remove the useful information,this article improves the threshold selection and threshold function which are the vital factors of this method.This article will improve the unified threshold to make the threshold vary with the changes of the decomposition scale.At the same time the deviation of wavelet coefficient and original coefficient will decrease,therefore,the method will retain the advantages of the traditional soft threshold and hard threshold.The proposed method with new threshold selection and threshold function will make it more flexible to deal with the wavelet coefficient in the process of image noise reduction with the help of wavelet decomposing.Experiment results with Matlab reveal that this technology can largely reduce the noise of the Camera image added with Gaussian noise,and the image contrast,S （entropy） and SNR （signal to noise ratio） also get improved.The effectiveness and precision of the new wavelet threshold method for image noise reduction presented in this article are finely proved by the results of the experiments.