小波阈值去噪法的深入研究

摘要

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 o-riginal 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.%针对以往小波阈值图像去噪法在去除图像噪声的过程中会出现的去噪不彻底、噪声残留、和噪声误判的问题,对小波阈值去噪方法中两个重要因素阈值选取方式和阈值函数进行改进,以达到更好去除噪声的目的.在以往的统一阈值基础上加以修改,使阈值能随着分解尺度的变化而改变,减少小波系数和原系数之间的偏差；对传统的软阈值和硬阈值的优点予以保留,改进它们各自的缺点,产生一种新的阈值函数,使它在处理小波系数时更加灵活.通过Matlab的仿真实验和对算法的精度分析表明,用改进后的小波阈值去噪法处理加高斯噪声的lean图像可以很好的去除图像噪声,使图像的信息熵,对比度和信噪比均得到很大的提高,图像质量和视觉效果也得到提升.