基于小波域隐马尔可夫树模型的医学图像去噪

Medical Image Denoising Based on Wavelet-Domain Hidden Markov Tree

目的:为了更好地去除DR医学图像噪声。方法:通过分析其噪声来源,在小波去噪和小波域隐马尔可夫模型的基础上,进行改进,即引入了方差不变性变换来调整原始图像的噪声模型为高斯噪声模型;图像分解为不同频率的不同子带,而隐马尔可夫树模型则用来规划小波系数的边缘分布。结果:自然图像处理实验结果表明,与普通的小波去噪方法相比,该方法不但可以保留图像的边缘信息,而且能提高去噪后图像的峰值信噪比。结论:同时用该方法处理DR图像,处理结果表明此方法在噪声去除、细节质量及骨骼锐化等方面比传统的高斯滤波及小波阈值滤波等方法效果要好。

Objective To denoise digital radiographic images well. Methods A technique was presented that used the Anscombe＇s transformation to adjust the original image to a Gaussian noise model based upon the wavelet denoising method and the wavelet-domain Hidden Markov Tree （HMT） model. Wavelet domain HMT models were used to determine the dependencies of multiscale wavelet coefficients through the state probabilities of the wavelet coefficients, whose sedistribution densities could be approximated by Gaussian mixture model. Results The proposed method could keep natural images edges from damaging and increase PSNR. Conclusion Quantitative and qualitative DR images assessment shows that the proposed algorithm outperforms the traditional Gaussian filter in terms of noise reduction, quality of details and bone sharpness.