摘要
针对侧扫声呐图像的斑点噪声,提出了一种贝叶斯估计的Curvelet域降斑方法。依据海底散射模型,得到侧扫声呐图像斑点噪声的瑞利分布乘性噪声模型。将取对数后的含斑图像进行Curvelet变换,依据噪声系数的瑞利分布、信号系数的高斯分布,结合贝叶斯理论的最大后验概率估计方法,在近似条件下,得到Curvelet变换域系数估计的理论解析式。采用局部自适应的邻域窗口确定方法,对Curvelet变换域处理后的系数进行逆变换,再经过指数变换后得到降斑的侧扫声呐图像。实验结果表明,在客观评价指标和主观视觉效果方面,新方法均取得了优于传统的空间滤波及基于小波的降斑方法的效果。
A curvelet domain method based on Bayesian estimation is proposed for side-scan sonar image despeckling.A multiplicative speckle noise model of Rayleigh distribution is established according to seabed scattering model.Sonar images are decomposed in curvelet domain after logarithmic transform.In curvelet domain,according to Rayleigh distribution of noise and Gaussian distribution of signal,the theoretical expression of coefficient estimation is obtained using maximum a posteriori estimation of Bayesian theory.Using local adaptive neighborhood window determination method,inverse curvelet transform is carried out to the coefficients of curvelet transform and then exponential transform is performed to obtain final side-scan sonar images.Experimental results show that the new method is superior to traditional spatial filtering methods and wavelet-based despeckling methods in terms of both objective evaluation and subjective visual effect.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2011年第1期170-177,共8页
Chinese Journal of Scientific Instrument
基金
国家自然科学基金(60972101
60872096)
疏浚技术教育部工程研究中心开放基金(HDCN08002)
中央高校基本科研业务费专项基金(2009B31814)资助
