对数域的光学相干层析图像噪声模型分析

Analysis of noise model of optical coherence tomography image in logarithmic domain

为了分析光学相干层析成像技术(OCT)的图像对数变换前后的斑点噪声分布模型,为图像去噪及其它图像处理过程提供参考,采用了先推导斑点噪声理论分布模型,再通过实验数据验证的方法。首先分析了对数变换前斑点噪声的统计特性,利用数学理论推导得到斑点噪声对数变换后的理论分布模型,然后用OCT心血管图像的平滑区域作为噪声的试验数据,得到变换前后斑点噪声的直方图分布数据,最后用理论分布模型对试验数据进行曲线拟合,取得了确定系数、均方根差和卡方检测的评价数据。结果表明,对数变换后斑点噪声分布符合FisherTippett分布。这一结果对OCT图像去噪等处理过程是有帮助的。

In order to analyze the speckle noise distribution model of optical coherence tomography （ OCT ） images before and after logarithm transformation and provide references for image de-noising and other image processing , theoretical distribution model of the speckle noise was deduced firstly and then was verified by experimental results .First of all, the statistical properties of the speckle noise before the logarithm transformation were analyzed .The distribution model of the speckle noise after logarithm transform was derived through the mathematical theory .Then, the smooth regions of cardiovascular OCT images were selected as the experimental data to obtain the histogram distribution of the speckle noise before and after the logarithm transformation .Finally, the experimental data were fitted by the theoretical distribution model and the evaluated data, such as R-square, root mean square error and X-square tests.The results prove that the noise distribution after logarithm transformation presents Fisher-Tippett distribution .The results are helpful for the de-noising on OCT images.