基于改进小波阈值函数的仪表图像去噪方法研究

Research of Instrument Image Denoising Based on Improved Wavelet Threshold Function

传统小波去噪的软阈值函数和硬阈值函数在去噪过程中产生的恒定偏差与抖动会导致去噪后的图像存在模糊和伪吉布斯现象。基于此问题,该文提出了一种软硬兼顾的加权平均阈值函数,此阈值函数避免了传统阈值存在的“一刀切”问题,降低恒定偏差的同时具有连续性与渐进性。其次在改进传统阈值门限上,设计了一种随着分解层数增加而减少的自适应门限阈值,避免消除深层数的高频边缘信息。最后通过对含有不同高斯白噪声的仪表图片进行去噪,计算峰值信噪比与均方误差并对结果图像进行的客观分析评估,实验结果表明,该阈值函数方法相比于其他方法,去噪结果更为清晰,均方误差最小,峰值信噪比最大,去噪效果更佳。

The soft threshold function and hard threshold function of traditional wavelet denoising will produce constant deviation and jitter in the denoising process,which will cause blur and pseudo-Gibbs phenomenon in the denoised result image.This paper studies the previous wavelet threshold denoising algorithm.,and proposes a weighted average threshold function that considers both soft and hard.This threshold function avoids the problem of“one size fits all”in traditional thresholds,and at the same time has continuity and asymptotics,reducing constant deviation.Secondly,it improves the traditional threshold,and designs an adaptive threshold that decreases with the increase of the number of decomposition layers,so as to avoid eliminating the high-frequency edge information of the deep layers.Finally,by denoising the instrument pictures with different Gaussian white noises,calculating the peak signal-to-noise ratio and the mean square error,the objective analysis and evaluation of the resulting images are carried out.The experimental results show that this threshold function method is better than other methods.Clear,the MSE is the lowest,the PSNR is the largest,and it has a better denoising effect.