Different resolution core three-dimensional image fusion method

Abstract

The application discloses a kind of different resolution core three-dimensional image fusion methods, it is proposed that the core image of different resolution from the same core is known, under the condition that low-resolution core image is enlarged uniform resolution by interpolation and ensures that its pore structure is unchanged, the equivalent spherical radius distribution of pore is calculated and counted, new high-resolution three-dimensional structure is obtained after processing, two high-resolution structures are scanned to establish pattern set, small pores in high-resolution structure are fused and reconstructed in low-resolution structure non-pore phase by the matching criterion, fusion formula and edge reconstruction modification criterion proposed.This application can fuse the core image of different resolution, reconstruct large-size core three-dimensional structure with large pore in low-resolution structure and small pore in high-resolution structure, and can be applied in later experimental analysis in the field of petroleum geology.

Description

Technical Field

[0001] This invention relates to a method for fusing three-dimensional core images of different resolutions, belonging to the field of three-dimensional image reconstruction technology. Background Technology

[0002] In the field of digital core imaging, there is often a trade-off between imaging field of view and imaging resolution. Scanning large-scale cores (typically on the centimeter scale) under low-resolution conditions results in a large field of view but low resolution, capturing only larger pores and failing to capture minute pores. Obtaining high-resolution core images requires cutting large core samples into smaller ones (typically on the millimeter scale). While this captures minute pores, the field of view shrinks. Due to the scale difference between large and small samples in real cores, direct stacking cannot effectively fuse pore information. Therefore, fusing pore information from core images of different resolutions using fusion reconstruction methods is essential.

[0003] Currently, in the field of digital cores, algorithms for 3D reconstruction under the same resolution conditions based on numerical reconstruction are relatively mature. However, there is still great potential for research in the direction of fusion reconstruction. Moreover, most of the current research on the fusion of images with different resolutions is based on the fusion of two-dimensional information, and there is relatively little research on fusion reconstruction directly using the three-dimensional information of pore morphology. Summary of the Invention

[0004] The main objective of this invention is to extract the morphological and positional distribution information of pores in a three-dimensional high-resolution core image, and then, without changing the pores in a low-resolution three-dimensional core image, fuse and reconstruct the pores in the high-resolution three-dimensional core image into the non-porous phase of the low-resolution structure, thereby obtaining a large-size three-dimensional high-resolution core image that simultaneously has macropores and micropores.

[0005] The present invention achieves the above objectives through the following technical solutions:

[0006] (1) Core images from the same core but with different resolutions: high-resolution small pore structures H(A) and H(B) and low-resolution macropore structure L;

[0007] (2) Determine a fixed-size three-dimensional template and scan the small hole structures H(A) and H(B) with a raster path to establish pattern set A and pattern set B accordingly;

[0008] (3) Use the three-dimensional template in step (2) to access the large-pore structure L in a random path to obtain the block T to be fused;

[0009] (4) Based on the matching criteria, obtain the best matching pattern P of the block T to be merged in step (3) from pattern set A or pattern set B.best ;

[0010] (5) According to the fusion formula, the block T to be fused in step (3) and the best matching pattern P in step (4) are fused together. best Perform fusion to obtain fused block R;

[0011] (6) Based on the edge matching and modification criteria, the fusion block R obtained in step (5) is reconstructed by edge modification;

[0012] (7) Repeat steps (3), (4), (5), and (6) until the entire three-dimensional structure reaches the preset porosity and the reconstruction is complete.

[0013] In step (1) of the above scheme, the macroporous structure L is the structure after magnification of the original low-resolution three-dimensional core image through pixel interpolation algorithm, and the microporous structure H(A) is the original high-resolution three-dimensional core image; the distribution of the equivalent sphere radius of the pores in the macroporous structure L and the microporous structure H(A) is statistically analyzed, and the microporous structure H(B) is the structure after removing the pores in the microporous structure H(A) that have the same equivalent sphere radius as the macroporous structure L; the formula for calculating the equivalent sphere radius of the pores in the core image is:

[0014]

[0015] Among them, V Pore The volume of the pore is represented by r, and the equivalent sphere radius of the pore is represented by r.

[0016] In the above scheme, the determination of the fixed-size three-dimensional template in step (2) is the length of the shortest line segment when the linear path function of the pore phase in the X, Y, and Z directions of the pore structure H(B) is zero, and the average value in the three directions is determined as the side length of the three-dimensional template; the linear path function L of the pore phase is... Pore (r) indicates that the endpoint is and The probability that line segment r belongs to the porous phase is defined as follows:

[0017]

[0018] The following relationship must be satisfied.

[0019]

[0020] in, For connecting endpoints and Any point on the line segment.

[0021] In the above scheme, the three-dimensional template of a fixed size mentioned in step (2) is defined as an N×N×N three-dimensional voxel block, and the number of voxels that the template can contain is N. 3 indivual.

[0022] In the above scheme, the grating path mentioned in step (2) is defined as follows: In a three-dimensional cuboid, a three-dimensional template is used to scan from left to right, then from front to back, and then from top to bottom, starting from the left front of the bottom layer of the cuboid. First, each position of the first row, first column, and first layer is scanned, then each position of the first row, first column, and second layer is scanned, and so on, until the entire three-dimensional cuboid is scanned.

[0023] In the above scheme, the specific operation of establishing the pattern set in step (2) is to encode the three-dimensional template scanning extraction pattern in binary encoding mode, sort it according to the number of holes contained in each pattern, assign the sorted patterns a number as the search index value, and store each encoded pattern and its index value as a whole in the corresponding pattern set.

[0024] In the above scheme, the random path mentioned in step (3) is defined as follows: in a three-dimensional cuboid, each point in the cuboid has a coordinate. A random algorithm is used to randomly generate a three-dimensional coordinate each time, and this coordinate is used as the position where the upper left vertex of the three-dimensional template will be placed.

[0025] In the above scheme, the matching criterion mentioned in step (4) is implemented as follows: The number of large holes in the obtained block T to be fused is counted. The number of large holes is defined as the number of voxel points of the pores in the large hole structure L contained in the template during this 3D template scan. If the number of large holes in the block is zero, a pattern is randomly selected from the pattern set B as the best matching pattern P for the block. best If the number of large holes in the block is greater than zero, according to the matching formula, the block to be fused T is compared with the patterns in the pattern set A in a finite or complete manner. The pattern with the largest matching degree (i.e., matching coefficient M) is the best matching pattern P. best The matching formula is defined as follows:

[0026]

[0027] Among them, M i For the matching coefficient, |F(P) i (x,y,z))*T(x,y,z)| represents the pattern P iThe absolute value of the product of the block to be fused and the block T at the corresponding position (x, y, z), where i is the number of the current pattern used to calculate the matching coefficient, t is the side length of the 3D template, and P best For the best matching pattern, F(P) i (x,y,z) is the pattern P i The three-phase shielding function used to calculate the matching coefficient is defined as follows:

[0028]

[0029] Wherein, P i (x,y,z) represents the pixel value of the i-th pattern at (x,y,z).

[0030] In the above scheme, the fusion formula mentioned in step (5) is defined as follows:

[0031]

[0032] Where R(x,y,z) represents the pixel value of the merged block R at position (x,y,z), T(x,y,z) represents the pixel value of the block to be merged T at position (x,y,z), and P... best (x,y,z) represents the optimal matching pattern P. best The pixel value at position (x,y,z).

[0033] In the above scheme, the edge matching modification criterion mentioned in step (6) is implemented as follows: After obtaining the fusion block R, it is detected whether the edge of the fusion block R contains reconstructed small hole points. For the edge containing hole points in the fusion block R, the three-dimensional template is translated along the direction of the edge and the block to be modified Q is obtained. According to the edge modification matching formula, the block to be modified Q is compared with the patterns in the pattern set in a limited or complete manner to find the pattern P that best matches the block to be modified Q (i.e., the similarity coefficient C is the largest). best And according to the fusion formula described in step (5), the block Q to be modified is matched with the best matching pattern P of edge modification reconstruction. best The blocks are merged, and the merged block becomes the new block to be modified. Edge modification and reconstruction are then performed on the new block until the edges of the block no longer contain hole points or are connected to large holes in that direction. This edge modification and reconstruction is then complete. The edge modification matching formula is defined as follows:

[0034]

[0035] Among them, C i For similarity coefficients, i represents the number of the pattern being calculated, t represents the side length of the 3D template, and P... i(x,y,z) represents the pixel value of the i-th pattern at (x,y,z), Q(x,y,z) represents the pixel value of the block Q to be modified at (x,y,z), and δ(·) is the Dirac function.

[0036]

[0037] P best The best matching pattern for the block Q to be modified. Attached Figure Description

[0038] Figure 1 This is a flowchart illustrating the implementation of the core image fusion method based on a dual-mode set at different resolutions in an example of the present invention.

[0039] Figure 2 The large-hole structure L (left), small-hole structure H (A) (middle), and small-hole structure H (B) (right) are shown in the examples of this invention.

[0040] Figure 3 This is a diagram showing the pore equivalent sphere radius distribution of the macropore structure L, micropore structure H(A), and micropore structure H(B) in the examples of this invention;

[0041] Figure 4 This is the result of fusion and reconstruction in the example of the present invention;

[0042] Figure 5 This is a comparison diagram of the fusion reconstruction results and the pore equivalent sphere radius distribution of the macroporous structure L in the example of this invention. Detailed Implementation

[0043] The present invention will now be described in more detail with reference to specific embodiments and accompanying drawings. However, the embodiments described are merely a specific and detailed description of the implementation method of the present invention and should not be construed as any limitation on the scope of protection of the present invention.

[0044] (1) The original three-dimensional low-resolution core image, with a size of 350×350×350 and a resolution of 10μm / pixel. Figure 2 The left image is the interpolated and magnified 3D low-resolution core image, i.e., macropore structure L, with dimensions of 700×700×700 and a resolution of 5μm / pixel; the middle image is the original 3D high-resolution core image, i.e., micropore structure H(A), with dimensions of 280×280×280 and a resolution of 5μm / pixel; the right image is the micropore structure H(B), with dimensions of 280×280×280 and a resolution of 5μm / pixel.

[0045] (2) Using formula (1), the equivalent sphere radius distribution range of the pores in the macroporous structure L and the microporous structure H(A) is statistically analyzed. The range is 24.01~278.79μm and 2.70~138.82μm, respectively. Then, in the microporous structure H(A), the pores with the pore radius distribution between 24.01~138.82μm are removed to obtain the microporous structure H(B). The pore diameter range of H(B) is 2.70~23.99μm. The equivalent sphere radius distribution of the pores of the three structures is shown in Table 1.

[0046] Table 1

[0047]

[0048] (3) Using formulas (2) and (3), the linear path function L of the pore phase in the X, Y, and Z directions of the small pore structure H(B) is statistically analyzed. Pore The shortest line segment length when (r) is zero is obtained by averaging the three values, which is 16, meaning the side length of the 3D template is 16.

[0049] (4) Use 16*16*16 three-dimensional templates to scan the hole structure H(A) and the hole structure H(B) respectively, and establish pattern set A and pattern set B accordingly.

[0050] (5) Using a 16*16*16 three-dimensional template, access the large-pore structure L via a random path. Based on the matching criteria and using formulas (4) and (5), search for the best matching pattern P of the block to be fused T. best .

[0051] (6) According to the fusion formula, use formula (6) to match the obtained block T to be fused with the best matching pattern P. best Perform fusion to obtain fused block R.

[0052] (7) Reconstruct the edges of the fused block R by matching the edge modification criteria and using formula (6).

[0053] (8) Repeat steps (5), (6), and (7) until the entire three-dimensional structure reaches the preset porosity of 0.0711, and the reconstruction is complete.

[0054] exist Figure 4 The image shows the fusion and reconstruction results of an example of the present invention. For easier observation, a 280*280*280 structure has been cropped for display. Figure 4 In the right figure, to more intuitively and quantitatively compare the accuracy of the fusion reconstruction results, the equivalent sphere radius distribution of the pores in the fusion reconstruction results and the macroporous structure is statistically analyzed. The comparison results are shown in the figure. Figure 5 In the example of this invention, the range of the equivalent sphere radius distribution of the pores in the fusion reconstruction result is the sum of that in the small-pore structure and the large-pore structure, and the specific data is shown in Table 2.

[0055] Table 2

[0056]

[0057] The above embodiments are merely preferred implementation examples of the present invention and are not intended to limit the technical solutions described in the present invention. Any technical solution that can be implemented based on the above implementation examples without creative effort should be considered to fall within the protection scope of the present invention.

Claims

1. A method for fusing three-dimensional images of rock cores at different resolutions, characterized by the following steps: (1) Core images from the same core but with different resolutions, high-resolution images showing small pore structures , and low-resolution large-pore structures ; (2) Determine a fixed-size 3D template and scan the hole structure using a raster path. , Establish corresponding pattern sets and pattern set ; (3) Use the three-dimensional template from step (2) to access the macroporous structure via a random path. Obtain the block to be merged ; (4) Based on the matching criteria, the obtained blocks to be merged are counted. The number of macropores contained in the block; if the number of macropores in the block is zero, a pattern is randomly selected from pattern set B as the best matching pattern for the block. If the number of large holes in the block is greater than zero, according to the matching formula, the block to be fused will be... The pattern is compared with the patterns in pattern set A in a finite or complete manner, and the pattern with the highest matching degree (i.e., matching coefficient M) is the best matching pattern. The matching formula is defined as follows: in, For matching coefficients, For pattern and the block to be merged In the corresponding position The absolute value of the product, This is the number of the pattern for which the matching coefficient is currently being calculated. The side length of the 3D template. For the best matching mode, For pattern The three-phase shielding function used to calculate the matching coefficient is defined as follows: in, For the first The pattern is in Pixel value at; (5) Based on the edge matching modification criteria, the fusion block obtained in step (5) is processed. Perform edge refinement and reconstruction; (6) Repeat steps (3), (4), (5), and (6) until the entire three-dimensional structure reaches the preset porosity and the reconstruction is complete.

2. The method for fusing three-dimensional core images of different resolutions according to claim 1, characterized in that... In step (2), a three-dimensional template of a fixed size is determined, and the hole structure is statistically analyzed. The length of the shortest line segment when the linear path function of the porous phase in the X, Y, and Z directions is zero is used to determine the average value in the three directions as the side length of the three-dimensional template.

3. The method for fusing three-dimensional core images of different resolutions according to claim 1, characterized in that... The fusion formula mentioned in step (5) is defined as follows: in, Indicates a fusion block exist Pixel value at the location, Indicates the block to be merged exist Pixel value at the location, Indicates the best matching pattern exist The pixel value at the location.

4. The method for fusing three-dimensional core images of different resolutions according to claim 1, characterized in that... The edge matching modification criterion mentioned in step (6) is implemented as follows: Obtain the fusion block. Then, inspect the fusion block. Does the edge contain reconstructed aperture points for the fusion block? The image contains the edges of the reconstructed pinholes. The 3D template is translated along the direction of these edges to obtain the block to be modified. The block to be modified is determined according to the edge matching modification formula. Perform a limited or full comparison with the patterns in the pattern set to find the block to be modified. The most matching pattern (i.e., the pattern with the highest similarity coefficient C) And according to the fusion formula described in step (5), the block to be modified is... Best matching pattern with edge refinement reconstruction The blocks are merged, and the merged blocks are the new blocks to be modified. The edge modification and reconstruction of the new blocks to be modified are continued until the edges of the blocks to be modified no longer contain the reconstructed small holes or are connected to the large holes in that direction. The edge modification and reconstruction is then complete. The above edge modification matching formula is defined as follows: in, The similarity coefficient, This indicates the number of the pattern currently used to calculate the similarity coefficient. This represents the side length of the 3D template. For the first The pattern is in Pixel value at that location, Block to be modified exist Pixel value at that location, It is the Dirac function. Block to be modified Best matching mode.

Images