基于脉冲耦合神经网络模型的小波自适应斑点噪声滤除算法

分析了维纳滤波原理和脉冲耦合神经网络(PCNN)模型的特点,根据斑点噪声统计模型的特征,结合小波变换方法,提出了一种基于PCNN模型的小波自适应斑点噪声滤除算法(W-PCNN-WD)来改善超声图像质量.首先,对超声图像进行对数变换,使斑点噪声转换为加性噪声；对医学图像进行维纳滤波处理,计算其加性噪声的标准方差,并以此作为小波阈值.然后,利用小波变换对图像进行预处理,利用PCNN在小波域中对小波系数进行相应的修正.最后,进行小波逆变换和指数变换,获得滤除噪声的图像.结果表明:本文提出的滤波方法优于其他滤波方法,当噪声方差为0.01时,本文滤波算法获得的峰值信噪比(PSNR)比经Wiener滤波方法获得的高出9 dB.该滤波方法能在有效去除超声斑点噪声的基础上保留图像的边缘细节信息,极大地改善了图像的视觉质量.

wavelet adaptive denoising method based on PCNN

The Wiener filtering principle and characteristics of a Pulse Couple Neural Network(PCNN) model were analyzed and a wavelet adaptive denoising method based on the PCNN(W-PCNN-WD)was proposed according to a statistical model of speckle noise combined with a wavelet transform to improve the quality of ultrasound image.Firstly,the ultrasound image was performed a log conversion to transform the speckle noise to an additive noise.Then,the Wiener filtering was used to process the medical image to get the variance of the additive noise as the threshold of wavelet.Furthermore,the image was preprocessed by the wavelet transform and wavelet coefficients were recomposed appropriately by using the PCNN.Finally,the image was processed again by the wavelet inverter and the exponential transforms to get a denoising image.The result shows that the proposed filtering method is better than the other filtering methods,and the Peak Signal to Noise Ratio(PSNR) from the proposed method is higher 9 dB than that from the Wiener filtering when the noise variance is 0.01.The method can keep the edge details of the information on the basis of removing speckle noise,which improves the visual quality of images greatly.