数据挖掘的光学断层成像重建方法

Optical tomography reconstruction method for data mining

为了有效重建光学断层成像,提升成像清晰度,研究数据挖掘的光学断层成像重建方法。依据光学断层成像原理,获取光学断层成像;利用改进模糊C均值聚类算法,划分光学断层成像图像块对类别;通过各类图像块对训练极限学习机回归器,估计高分辨率图像块对垂直梯度、方向梯度和高频;以垂直梯度、方向梯度为约束,重建光学断层成像;以高频与重建光学断层成像为约束,再次重建光学断层成像,得到最终重建光学断层成像。实验证明:该方法可以有效重建模拟与真实的光学断层成像,提升光学断层成像清晰度;精准聚类光学断层成像图像块对;在不同图像块尺寸时,重建光学断层成像的位置误差与质心误差均较低,图像相似度较高,具有较高的重建精度。

In order to effectively reconstruct optical tomography and improve imaging clarity,the optical tomography reconstruction method based on data mining is studied.According to the principle of optical tomography,optical tomography is obtained;The improved fuzzy c-means clustering algorithm is used to classify the block pairs of optical tomography images;By training the extreme learning machine regressors with various image block pairs,the vertical gradient,directional gradient and high frequency of high-resolution image block pairs are estimated;Constrained by vertical gradient and direction gradient,optical tomography was reconstructed;With high frequency and reconstruction optical tomography as constraints,optical tomography is reconstructed again to obtain the final reconstruction optical tomography.Experiments show that this method can effectively reconstruct simulated and real optical tomography,and improve the clarity of optical tomography;Accurate clustering of optical tomography image block pairs;At different image block sizes,the position error and centroid error of reconstructed optical tomography are low,the image similarity is high,and the reconstruction accuracy is high.