摘要
从图像的去噪模型入手,引入基于偏微分方程(PDE)的正则空间模型,结合全变差(TV)滤波器的设计,给出了一种针对声纳图像去噪的方法及其实现,并提出了基于小波变换的噪声方差估计方法。结果表明,由于采用不同于最小均方误差准则的新准则函数,在保持方差不变的条件下利用图像梯度信息建立选择性异性扩散模型来进行图像去噪复原,从而达到了既保护图像边缘又去除噪声的目的;与基于软阈值的小波去噪方法相比,在峰值信噪比和边缘保留评价参数方面具有优势。
Started with the image denosing model, a PDE - based regularity spaces model was introduced. Combined with the idea of TV filter, a method for sonar image denoising was presented with its realization, and a wavelet - based noise variance estimation method was proposed. According to experiment, because we using a new criterion function to construct a selective anisotropy diffusion model for image denoising, which is different from the old one based on minimum mena square error（ MMSE ）, this method can holding the image edge while denoising it, and has higher signal to noise ratio and edge preserved index compared with the soft threshold based wavelet denoising.
出处
《黑龙江大学自然科学学报》
CAS
北大核心
2008年第1期10-14,共5页
Journal of Natural Science of Heilongjiang University
基金
国家自然科学基金资助项目(60672034)
高等学校博士学科点基金资助项目(20060217021)
黑龙江省自然科学基金重点项目(ZJG0606-01)
关键词
声纳图像
图像去噪
偏微分方程
TV滤波器
全变差模型
小波去噪
sonar image
image denoising
partial differential equations
TV filter
total variation model
wavelet denoising
