基于模拟退火的多尺度岩心三维图像融合重建

Three-dimensional image fusion reconstruction of multi-scale core based on simulated annealing

摘要:

在岩心图像分析中，低分辨率岩心图像能够显示较大视域，具有较好的全局代表性，而对于低尺度信息无法准确表征；高分辨率岩心图像能够准确地表征岩心的低尺度信息，但通常仅能显示较小的视域.为了能综合分析高低分辨率下的不同岩心图像，本研究提出了一种基于模拟退火，将低分辨率三维岩心图像和高分辨率二维岩心图像融合重建为高分辨率三维岩心的算法.具体地，对于给定的高分辨率二维岩心图像，首先，将低分辨率三维岩心图像进行插值放大以统一两者的点长度，统计二者在二维中的孔隙分布情况，只保留高分辨率二维岩心图像中的小尺寸孔隙，以作为训练图像；然后，在融合重建过程中将低分辨率三维岩心中的大尺寸孔隙相设置为硬数据，以两点相关函数为目标函数重建其中的小尺寸孔隙.实验结果表明，本研究提出的融合重建算法可以很好的将低分辨率岩心重建为高分辨率岩心结构，且融合重建结果有效，准确.

Abstract:

In core image analysis, Lowresolution core images can display a larger fields of view and have better global representation, but cannot accurately characterize the lowscale information; Whilehighresolution core images can accurately represent the lowscale information of the core,but usually only show a smaller field of view. In order to comprehensively analysis different core images in high and low resolution, An algorithm is proposed based on simulated annealing to reconstruct lowresolution 3D core image into highresolution 3D core with highresolution twodimensional (2D) core image being the constraint information. For a given highresolution 2D core image, the lowresolution 3D core image is first interpolated to unify the length of the any two points, the pore distribution in the 2D core image is counted, and the small pores in the highresolution 2D image are retained as a training image. In the fusion reconstruction process, the largesize pores in the lowresolution 3D core are set as hard data, and the autocorrelation function is used as the objective function to reconstruct the smallsized pores. The experimental results show that the fusion reconstruction algorithm proposed in the paper can reconstruct the lowresolution core into a highresolution core structure, and the fusion reconstruction results are effective and accurate.