Multiscale modeling algorithm for core images

Computed tomography (CT) images of large core samples acquired by imaging equipment are insufficiently clear and ineffectively describe the tiny pore structure; conversely, images of small core samples are insufficiently globally representative. To alleviate these challenges, the idea of a super-resolution reconstruction algorithm is combined with that of a three-dimensional core reconstruction algorithm, and a multiscale core CT image fusion reconstruction algorithm is proposed. To obtain sufficient image quality with high resolution, a large-scale core image is used to provide global feature information as well as information regarding the basic morphological structure of a large-scale pore and particle. Then the texture pattern and the tiny pore distribution information of a small-scale core image is used to refine the coarse large-scale core image. A blind image quality assessment is utilized to estimate the degradation model of core images at different scales. A multilevel pattern mapping dictionary containing local binary patterns is designed to speed up the pattern matching procedure, and an adaptive weighted reconstruction algorithm is designed to reduce the blockiness. With our method, images of the same core at different scales were successfully fused. The proposed algorithm is extensively tested on microstructures of different rock samples; all cases of the reconstructed results and those of the actual sample were found to be in good agreement with each other. The final reconstructed image contains both large-scale and small-scale information that can provide a better understanding of the core samples and inform the accurate calculation of parameters.

Figure 1

Multiscale images of the same sample: (a) large-scale image at $\mathrm{\Phi }=25\mathrm{mm}(14\mu \mathrm{m}/\mathrm{pixel})$ and (b) small-scale image at $\mathrm{\Phi }=2\mathrm{mm},0.875\mu \mathrm{m}/\mathrm{pixel}$.

Figure 2

Core sample and degradation images with different degradation model and parameters: (a) subimage of the large-scale image; (b) subimage of the small-scale image; (c) down-sampled image of (b); (d) degraded image no. 1: (c) with Gaussian filtering ($\sigma =1.6$, kernel size $=9\times 9$); (e) degraded image no. 2: (c) with zero mean, Gaussian white noise with variance of 0.1 added followed by mean filtering (kernel size $=9\times 9$); (f) degraded image no. 3: (c) with zero mean, Gaussian white noise with variance of 0.1 added followed by Gaussian filtering ($\sigma =1.6$, kernel size $=9\times 9$); (g) degraded image no. 4: (c) with zero mean, Gaussian white noise with variance of 0.1 added followed by mean filtering (kernel size $=12\times 12$); (h) degraded image no. 5: (c) with zero mean, Gaussian white noise with variance of 0.1 added followed by Gaussian filtering ($\sigma =1.6$, kernel size $=12\times 12$); (i) degraded image no. 6: (c) with mean filtering (kernel size $=19\times 19$); (j) degraded image no. 7: (c) with zero mean, Gaussian white noise with a variance of 0.1 added followed by mean filtering (kernel size $=20\times 20$).

Figure 3

Overview of up-sampled image and small-scale image: (a) Up-sampled image of the selected degraded image and (b) small-scale image.

Figure 4

Pattern augmentation.

Figure 5

T pattern and its pixel neighborhood.

Figure 6

Pattern and its LBP.

Figure 7

Pattern mapping dictionary and its generation method.

Figure 8

Diagram of the SSIM system [27].

Figure 9

Fusing procedure framework: (a) Pattern mapping dictionary generation and (b) simulation.

Figure 10

Image of rock sample, down-sampled image, enlarged image, and pore size distributions: (a) original image of rock sample; (b) degraded image of (a) from a mean filter with a 5 × 5 kernel size and a down-sample factor of 8; (c) enlarged image of the degraded image (b); and (d) pore size distributions of (a) and (b).

Figure 11

(a) Small-scale image, (b) degraded image, and (c) enlarged image of (b).

Figure 12

Fusing results and comparisons: (a) original image, (b) degraded image, (c) enlarged image of (b), (d) fusing result, and (e)–(g) corresponding binary images.

Figure 13

Pore size distributions of the original images and the fused image.

Figure 14

Pore size distributions of ten reconstruction results with the same sample.

Figure 15

Core samples for simulation: (a) large-scale image with $\mathrm{\Phi }=25\mathrm{mm}$, 5 μm/pixel and (b) small-scale image with $\mathrm{\Phi }=5.7\mathrm{mm}$, 2.6 μm/pixel.

Figure 16

Multiresolution images of rock sample and reconstructed results: (a) small-scale image, (b) large-scale image, and (c) reconstructed result.

Figure 17

Binary images of (a) the large-scale image and (b) the reconstructed result.

Figure 18

Comparison of pore size distributions.

Figure 19

Pore size distributions of ten reconstruction results with the same sample.

Figure 20

Overview of rock sample and reconstructed results: (a) Large-scale 3D image; (b) small-scale 3D image enlarged twice; and (c) reconstructed result. Visual comparison of slices of the large-scale 3D image and reconstructed result: (d) tenth slice of the large-scale 3D image; (e) tenth slice of the reconstructed 3D image; (f) 50th slice of the large-scale 3D image; (g) 50th slice of the reconstructed 3D image; (h) 100th slice of the large-scale 3D image; (i) 100th slice of the reconstructed 3D image; (j) 150th slice of the large-scale 3D image; (k) 150th slice of the reconstructed 3D image; (l) 200th slice of the large-scale 3D image; and (m) 200th slice of the reconstructed 3D image.

Figure 21

Comparison of pore size distributions.