Abstract: A new wavelet threshold function is proposed for images of strong Gaussian noise. The traditional soft threshold de-noising method has a significant effect, but not ideal effect of strong Gaussian noise. In the soft threshold function, this paper constructs a new wavelet threshold function；the function contains the threshold value[λ], adjustable factor [t] and [n], which can adjust to the adaptive threshold changes. The experiment adopts the noised image and de-noised image Peak Signal to Noise Ratio（PSNR） criterion of maximize, based on Particle Swarm Optimization（PSO） using function selection parameters. The simulation results show that this method can not only effectively remove the noise, but also to avoid the loss of high frequency information useful to improve the signal to noise ratio. Especially in the strong Gaussian noise, the proposed method can relatively improve PSNR 6~7 dB than soft threshold method, it shows that this proposed threshold method can effectively reduce strong Gaussian noise on images.

Key words: wavelet transform, threshold, image de-noising, PSO algorithms

摘要： 针对图像中的强高斯噪声提出了一种新的小波阈值降噪函数。传统的软阈值法对图像去噪有明显的效果，但对强高斯噪声效果不甚理想，于是构造出一种新的小波阈值函数，此函数包含阈值[λ]，调节因子[t]和[n]三个参数，能够自适应地调节阈值的变化。实验以噪声图像与去噪后图像之间的峰值信噪比（PSNR）最大化为准则，采用PSO粒子群算法优化阈值函数中参数[n]和[t]的选取。仿真实验结果表明该方法不仅可以有效地去除噪声，又能避免有用高频信息的损失，提高了图像的信噪比；尤其在强高斯噪声下，相对软阈值法PSNR可提高6~7 dB，表明了此改进阈值法对于强高斯噪声图像降噪的有效性。

关键词: 小波变换, 阈值法, 图像去噪, PSO算法