摘要
本文提出了一种基于量子衍生参数估计的医学超声图像去斑方法.通过对对数变换的超声图像小波系数建模,提出了一种带自适应参数的概率分布函数.该方法充分考虑了小波系数的尺度间相关性,利用父-子代小波系数的归一化乘积,首次在高频子带中引入量子衍生信号与噪声出现概率.并利用贝叶斯估计理论,提出了一种基于量子衍生参数估计的自适应收缩函数.实验结果表明本方法较相关算法具有更好图像细节保持能力,去斑效果显著.
A novel despeckling method for medical ultrasound images is proposed based on quantum-inspired parameters estimation.A new probability distribution function with an adaptive parameter is built up with the modelling of log-transformed medical ultrasound images coefficients.Considering the inter-scale dependency of coefficients,the quantum-inspired probability of signal and noise is firstly introduced based on the normalized products of the coefficients and their parents.Using the Bayesian estimation theory,an adaptive shrinkage function is proposed based on quantum-inspired parameters estimation.Experiments showed that our method can notably reduce speckle noise and preserve details of medical ultrasound image effectively,which achieved much better performance than that of the other related despeckling methods.
出处
《电子学报》
EI
CAS
CSCD
北大核心
2011年第4期812-818,共7页
Acta Electronica Sinica
基金
国家自然科学基金(No.60672057
60804031)
国家863技术研究发展计划(No.2007AA12Z166)
中国科学院模式识别国家重点实验室开放课基金
关键词
去斑
量子信号处理
双树复小波变换
despeckling
quantum signal processing
dual-tree complex wavelet transform