Rock core information acquisition method based on close-range photography

Abstract

The invention discloses a rock core information acquisition method based on close-range photography. The method comprises the following steps: arranging a plurality of targets on a base; the rock core is fixed to the base through the supporting column, and the rock core is polished through the lighting system; shooting is rapidly completed at the fixed radius around the rock core according to the pitch angle, shooting is carried out again according to the next pitch angle till shooting is completed at all the pitch angles, and multiple sets of photos are obtained; respectively marking target center points and numbers on the plurality of groups of photos according to the target coordinates; obtaining image external parameters and sparse point clouds of an object space; creating a dense point cloud; generating a three-dimensional grid of the rock core according to the dense point cloud; mapping the color of the original image to the surface of the three-dimensional grid and the dense point cloud to form a color texture point cloud; and forming a rock core expansion image map according to the three-dimensional grid and the color texture point cloud of the rock core. The target layout, control measurement and calculation method provided by the invention is relatively high in operability and relatively high in control measurement precision, and relatively high precision is established from the initial stage of rock core information acquisition.

Description

Technical Field

[0001] The present invention relates to a method for collecting core information based on close-range photography, belonging to the technical field of geological exploration. Background Art

[0002] Physical cores are large in quality and volume, with high storage and transportation costs, difficult information retrieval, weak data extraction and sharing, which limit the scope and depth of analysis and research. In addition, after long-term preservation, weathering has caused earth-shaking changes in the color, mineral composition, water content, etc. of the cores, making it impossible to reflect the actual situation of the original state. By various technical means, converting physical cores into information such as text, pictures, and data that can be recognized, stored, and analyzed by computers, and converting the storage of physical cores into an "electronic digital core library" for electronic storage, is the best means to achieve long-term preservation, diverse services, wide sharing, information mining, and value growth of core information. Currently, there are more than 10 methods for obtaining core information data, which can be divided into 3 categories: surface image, chemical parameter, and physical parameter information collection. However, most of the technical equipment is of high value, has many supporting devices, poor portability, and is deployed in laboratories, increasing the core transportation cost. At present, there is a need for a method and system for collecting core information at the drilling site.

[0003] With the improvement of the capabilities and the reduction of the costs of photographic equipment, and the rapid development of computer technology, close-range photography has been widely used in various fields of the geological industry. Therefore, when collecting core information based on close-range photography technology, it is necessary to comprehensively consider the field working conditions such as drilling, investigation, and research, and integrate multidisciplinary technologies such as comprehensive collection, processing, matching, and unfolding to form specific technical characteristics and requirements. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for collecting core information based on close-range photography in view of the problems existing in the prior art. The present invention can form colorized dense point clouds, three-dimensional meshes, and two-dimensional unfolded diagrams, and can obtain the most intuitive and high-quality core surface images, which helps to analyze and extract parameters such as cracks, geometric parameters, structural types, quality indicators, minerals, and lithologies in the core.

[0005] The technical solution provided by the present invention to solve the above technical problems is: A method for collecting core information based on close-range photography, comprising the following steps: S1. Set a plurality of target marks on the base and determine the target mark coordinates; S2. Fix the core on the base through a support column and illuminate the core through an illumination system; S3. Surround the core at a certain pitch angle, quickly complete the shooting at a fixed radius, then shoot again at the next pitch angle until all pitch angles are completed, and obtain multiple groups of photos; S4. Mark the center points and numbers of the targets on multiple groups of photos according to the target coordinates; S5. Obtain the external parameters of the images and the sparse point cloud in the object space through aerial triangulation; S6. Create a dense point cloud based on the sparse point cloud in the object space; S7. Generate a three-dimensional mesh of the core according to the dense point cloud; S8. Map the color of the original image to the surface of the three-dimensional mesh and the dense point cloud to form a colored texture point cloud; S9. Form a core unfolded image according to the three-dimensional mesh of the core and the colored texture point cloud.

[0006] A further technical solution is that the target coordinates in step S1 are obtained through the following steps: S11. Number the targets and measure the distances between each target using a vernier caliper; S12. Set up a coordinate system and calculate the approximate coordinates of other unknown points using the principle of side intersection and the distances between each target; S13. Calculate the adjusted coordinates of all points using the principle of indirect adjustment.

[0007] A further technical solution is that the calculation formulas in step S12 include:

[0008]

[0009]

[0010]

[0011]

[0012]

[0013] In the formula: 、 、 and are the coordinates of the known points A and B ; is the distance between the known points A and B; and are the approximate coordinates of the unknown point D ; L 1 and L 2 are the distances from the points A 、 B to the point D respectively; l isL 1 Projected on the AB side, h is the height perpendicular to the AB side.

[0014] A further technical solution is that the calculation formula in step S13 is:

[0015]

[0016]

[0017] In the formula: and are the adjusted coordinates of point i ; and are the approximate coordinates of point i ; and are the least squares equation parameters; l 1 is the error equation constant; B is the design matrix; P is the weight matrix; T is the transposed matrix.

[0018] A further technical solution is that the specific process of step S5 is: Based on the collinearity condition equation, through the image point observations on multiple images, jointly solve the exterior orientation elements of the images and the three-dimensional coordinates of the object points, and jointly control the points through least squares adjustment to solve the exterior parameters of the images and form a sparse point cloud of the object space.

[0019] A further technical solution is that the specific process of step S6 is: Generate the depth value of each pixel through pixel-level matching of multi-view images, optimize the depth map based on semi-global matching, convert each pixel to the coordinates in the camera coordinate system in combination with the camera internal parameters, and project them to the global coordinate system through the camera external parameters. Finally, improve the point cloud quality through filtering, fusion, and isolated post-processing steps, and fuse them into a dense point cloud.

[0020] A further technical solution is that the specific process of step S7 is: Further denoise, filter, and resample the dense point cloud by removing outliers, isolated points with insufficient number of points in the neighborhood, or duplicate points whose distance exceeds 3σ from the mean; then calculate the point cloud normal vector based on the KD-Tree neighborhood, construct an octree structure, and convert the point cloud into an indicator function gradient field; and solve the Poisson equation to extract the isosurface to generate the three-dimensional mesh of the core.

[0021] A further technical solution is that step S8 includes the following steps: S81. Determine the visibility of each mesh patch or vertex based on the depth map; S82. Sample colors from the visible images using bilinear interpolation and multi-view color fusion, and fuse them onto the 3D mesh surface and point cloud; S83. Project three identical textures or different textures onto the model surface along three axes; according to the surface normal direction, perform weighted mixing on the three projection results to ensure natural texture transition; and perform final color calculation to balance the brightness and color difference of different images, and finally generate the core surface texture.

[0022] A further technical solution is that the calculation formula in step S83 includes:

[0023]

[0024]

[0025] In the formula: P x , P y , P z are the two-dimensional coordinates of the 3D point on the projection planes of each axis; t is the texture repetition times; o is the texture translation offset; w x , w y , w z are the weight coefficients of the three axial directions; n is the surface normal vector; X axis , Y axis , Z axis are the unit vectors of the three axial directions; k is the sharpness factor; T x , T y , T z are the texture sampling functions of the three axial directions.

[0026] A further technical solution is that the specific process of step S9 is: when cutting the 3D mesh and colored texture point cloud of the core along a vertical line and unfolding the core cylinder into a plane rectangle, each point on the core surface is mapped to a point on the plane to form a core unfolded image.

[0027] Advantages of the present invention: The core information acquisition device of the present invention is very simple, economical and easy to move. The acquisition of materials is extremely easy, and it has strong adaptability to field work. The proposed target layout, control measurement and calculation methods are highly operable, and the control measurement accuracy is relatively high, laying a high-precision foundation from the initial stage of core information acquisition. More detailed regulations are made for photo acquisition, reducing the influence of human factors and the difficulty of subsequent processing. The six steps of photo data processing are clear, the formula expressions are clear, and the formation of colored dense point clouds, three-dimensional meshes and two-dimensional surface models can intuitively reflect the surface characteristics of the core. Description of the Drawings

[0028] Figure 1 Schematic diagram of the layout for core photography; Figure 2 Target layout diagram for core photography; Figure 3 Calculation diagram of side length intersection; Figure 4 Schematic diagram of the core photography angle; Figure 5 Colored dense point cloud data of the core; Figure 6 Core color texture mapping and unfolding diagram. Detailed Implementation Modes

[0029] The technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0030] A core information acquisition method based on close-range photography provided by the present invention uses the placement system as shown in Figure 1 to place the core. The core placement system includes three parts: a base, a support column and a light. The base is a 5-mm thick steel plate to ensure the stability of the system during photo shooting. The support column is a steel bar with a diameter of 2 cm and a height of 5 cm, and the top is a tray with a diameter of about 5 cm. Clamping buckles are designed around it to maintain sufficient supporting force and fixing force for the core. Lights with a color rendering index (Ra) greater than or equal to 98 and a fixed color temperature without stroboscopic are selected and evenly arranged around the core. A soft light cover or a diffused light source is used to ensure that the core is evenly illuminated and avoid direct strong light. Specifically, it includes the following steps: S1. Set several targets on the base. The targets can be black and white checkerboards, QR codes, or circular codes, avoiding reflective materials. The targets should be evenly distributed around the support columns to avoid local concentration, with different spacings and angles. Ensure that at least 4 targets are visible in each photo, and the number exceeds 20% of the minimum requirement in key areas such as crack development, abnormal color, sunken pits, and particle changes. The target size should be clearly visible in the photo, preferably a square or rectangle with a side length of 10 - 20 cm. And determine the target coordinates through the following steps; S11. Number the targets according to S1 (left), CAL - 110 (right) in Figure 2 or other reasonable methods, and measure the distances between each target using a vernier caliper, such as Figure 2 D26 (left) or the red line (right) in S12. Set up a coordinate system, with point A as the origin (0, 0, 0) and point B as the known point. Using the principle of side - length intersection ( Figure 3 shown), starting from the known points A and B, and the distances between each target, calculate the approximate coordinates of other unknown points using equations 1 - 6; (1) (2) (3) (4) (5) (6) In the formula: , , and are the coordinates of the known points A and B ; is the distance between the known points A and B; and are the approximate coordinates of the unknown point D ; L 1 and L 2 are the distances from the points A , B to the point D respectively; l is L 1 projected on the AB side, h is the height perpendicular to the AB side; S13. Adopt the principle of indirect adjustment, construct a normal equation according to the least - squares principle as shown in equation 7, and solve the parameters of the normal equation and calculate the adjusted coordinates of all points using Equations (8) and (9); (7) (8) (9) wherein: and are the adjusted coordinates of point i ; and are the approximate coordinates of point i ; and are the parameters of the least squares equation; l 1 is the constant of the error equation; B is the design matrix; P is the weight matrix; T is the transposed matrix; S2. Fix the core on the base through the support column, and ensure that no other objects block the core, avoiding transparent and reflective objects; illuminate the core through the lighting system to ensure consistent lighting conditions; S3. Rotate around the core at a certain pitch angle, and quickly complete the shooting at a fixed radius. Then, shoot again at the next pitch angle until all pitch angles are completed, obtaining multiple groups of photos; S31. Shoot around the core from multiple horizontal and vertical angles, while ensuring that there is a 60 - 80% overlap between the left and right in the horizontal direction, and shoot in a circle at an angle of 10 - 30°; in the vertical direction, there is a 30 - 50% overlap rate required, and shoot at multiple angles such as 20 - 45° from the top view, 20 - 45° from the front view and 20 - 45° from the bottom view ( Figure 4 ); take backup photos for key areas; the shooting range covers at least 4 or more targets; S32. Do not turn on the flash of the camera; fix the focal length, maximum resolution, as low ISO as possible, a smaller aperture, a fast shutter speed, and do not set automatic white balance; try to select the RAW format for the photos and convert them to TIFF data later; use a tripod and a shutter release cable to ensure that the photos are not blurred; S33. The core cannot move; rotate around the core at a certain pitch angle, and quickly complete the shooting at a fixed radius. Then, shoot again at the next pitch angle until all pitch angles are completed; S34. Immediately organize the photos after shooting, check the clarity of the photos, the visibility of the targets, and the unevenness of the lighting, eliminate unqualified photos and immediately reshoot; S35. After the photos are checked and qualified, repeat S31 to S33; S4. Mark the center points and numbers of the targets on multiple groups of photos according to the target coordinates; S5. Obtain the exterior parameters of the images and the sparse point cloud in the object space through aerial triangulation; Based on the collinearity condition equation (Equation 10), through the image point observations on multiple images, jointly solve the exterior orientation elements (position and attitude) of the images and the three-dimensional coordinates of the object space points, and through least squares adjustment, jointly with the control points to achieve global optimization, solve the exterior parameters of the images and form the sparse point cloud in the object space; (10) In the formula: and are the projection center coordinates (exterior orientation line elements) of the i th image, represented by the translation vector T ; w i , Φ i , k i are the attitude angles (exterior orientation angle elements) of the i th image, represented by the rotation matrix R i = , and ( k= 1,2,3 ); 、 and f are the principal point coordinates and focal length (interior orientation elements) of the i th image; 、 and are the three-dimensional coordinates of the j th object space point; and are the image point coordinates of the j th object space point on the i th image; and are the observation errors (or residuals) of the image point coordinates; S6. Create a dense point cloud based on the sparse point cloud in the object space; Through pixel-level matching of multi-view images, generate the depth value of each pixel, optimize the depth map based on semi-global matching, convert each pixel to the coordinates in the camera coordinate system in combination with the camera internal parameters (Equation 11), and project it to the global coordinate system through the camera external parameters (Equation 12). Finally, improve the quality of the point cloud through post-processing steps such as filtering, fusion, and isolation, and fuse it into a dense point cloud ( Figure 5 ); (11) (12) In the formula: P is the point in the global coordinate system, X, Y, Z is the global coordinate, X c 、Y c 、Z c is the coordinate of the depth image pixel in the camera coordinate system; d is the depth value of the depth image pixel; u, v is the coordinate of the depth image pixel (with the upper left corner of the image as the origin); f is the focal length of the camera (in pixels); x 0 、y 0 is the principal point coordinate; S7. Generate a three-dimensional mesh of the core according to the dense point cloud; By deleting the points whose distance exceeds 3 σ Remove outliers, isolated points with insufficient points in the neighborhood or duplicate points, and further perform denoising, filtering, resampling, etc. on the dense point cloud. Calculate the normal vector of the point cloud based on the KD-Tree neighborhood, construct an octree structure, and convert the point cloud into an indicator function gradient field. Solve the Poisson equation (Equation 13), extract the isosurface to generate a three-dimensional mesh. Interpolate new patches along the hole edges based on curvature to fill the holes, and reduce noise based on Laplacian smoothing; (13) In the formula: is the implicit function (1 inside the object and 0 outside); is the gradient of the indicator function (non-zero at the surface, and the direction is the normal vector); is the divergence of the point cloud normal vector field, reflecting the surface curvature (positive divergence in convex regions and negative in concave regions); S8. Map the color of the original image to the surface of the three-dimensional mesh and the dense point cloud to form a colored texture point cloud; S81. Determine the visibility of each mesh patch or vertex based on the depth map; Project the vertices of the three-dimensional mesh P onto the image pixel coordinate system P c (Equations 14 and 15). If the depth P after projection of the vertex of the three-dimensional mesh P z and the depth value u, v of the image depth map at this pixel ( d are within a certain tolerance threshold (Equation 16), then it is considered that this vertex of the three-dimensional mesh is visible in this image; (14) (15) (16) In the formula: P w is the coordinate of the 3D mesh vertex P in the world coordinate system; P c,x 、 P c,y and P c,z are respectively the P coordinates of the 3D mesh vertex X 、 Y and Z in the camera coordinate system; ε is the tolerance threshold for visibility discrimination; S82. Sample colors from visible images using bilinear interpolation (Equation 17) and multi-view color fusion (Equation 18), and fuse them onto the 3D mesh surface and point cloud; (17) (18) In the formula: I ( u, v ) is the color value of the image at pixel coordinates ( u, v ); and are floating-point numbers u and v are the rounded-down integers; w ij is the bilinear interpolation weight, determined by the distance between the sub-pixel position and neighboring pixels; C final is the finally fused color of the vertex; w i is the weight of the i th view, based on the view angle or distance; S83. Project three identical textures or different textures onto the model surface along three axes according to Equation 19; according to the surface normal direction, perform weighted mixing on the three projection results according to Equation 20 to ensure natural texture transition; perform the final color calculation according to Equation 21 to balance the brightness and color difference of different images, and finally generate the core surface texture ( Figure 6 ); (19) (20) (21) In the formula: P x 、P y , P z are the two-dimensional coordinates of the three-dimensional point on the projection planes of each axis, such as P x = ( y, z ); t is the texture repeat count; o is the texture translation offset; w x , w y , w z are the weight coefficients of the three axes; n is the surface normal vector; X axis , Y axis , Z axis are the unit vectors of the three axes, such as X axis =(1, 0, 0); k is the sharpness factor, usually 2 - 4, and the larger the value, the narrower the mixing region; T x , T y , T z are the texture sampling functions of the three axes; S9. Form the core unfolded image according to the three-dimensional grid and the colored texture point cloud of the core; When cutting the core cylinder along a vertical line ( θ = 0) and unfolding it into a plane rectangle, each point ( θ , z ) on the core surface is mapped to the point ( x , y ) on the plane according to Equation 22, forming the core unfolded image ( Figure 6 ); (22) In the formula: x is the abscissa of the core surface unfolding, corresponding to the bottom circumference of the core, with a range of [0, 2 πr ); y is the ordinate of the core surface unfolding, corresponding to the core height, with a range of [0, h ); θ is the angle around the core for one week, with a range of [0, 2 π ); z is the core height value, with a range of [0, h ); r is the core radius.

[0031] As mentioned above, this is not any form of limitation to the present invention. Although the present invention has been disclosed through the above embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to equivalent embodiments of equivalent changes by using the disclosed technical content within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A method for collecting core information based on close-range photography, characterized in that: The following steps are involved: S1. Set several targets on the base and determine the target coordinates; S2, fix the core on the base through the support column, and illuminate the core through the lighting system; S3, quickly complete the shooting with a fixed radius around the core according to the pitch angle, and then press the next pitch angle to shoot again until all pitch angles are shot, and obtain multiple sets of photos; S4, marking the target center points and numbers on multiple groups of photos according to the target coordinates; S5, obtaining the sparse point cloud of image external parameters and object space through aerial triangulation; S6, creating a dense point cloud based on the sparse point cloud of the object space; S7, generating a three-dimensional grid of the core according to the dense point cloud; S8, mapping the original image color to the three-dimensional mesh surface and the dense point cloud to form a color texture point cloud; S9. A core unfolding image is formed based on the three-dimensional grid and color texture point cloud of the core.

2. A method for collecting core information based on close-range photography according to claim 1, characterized in that: The target coordinates in step S1 are obtained by the following steps: S11, numbering the targets and measuring the distance between each target using a vernier caliper; S12, setting a coordinate system, and calculating the approximate coordinates of other unknown points using the edge length intersection principle and the distances between targets; S13. Use the indirect adjustment principle to calculate the adjustment coordinates of all points.

3. A method for collecting core information based on close-range photography according to claim 2, characterized in that: The calculation formula in step S12 includes: Where: , , and For known points A and B The coordinates of is the distance between known points A and B; and For unknown points D The approximate coordinates of L 1 and L 2 are points A , B To point D distance; l for L 1 in AB Edge projection, h is perpendicular to AB The height of the side.

4. The method for collecting core information based on close-range photography according to claim 2, characterized in that: The calculation formula in step S13 is: Where: and For point i The adjusted coordinates of and For point i The approximate coordinates of and are the least squares equation parameters; l 1 is the error equation constant; B is the design matrix; P is the weight matrix; T is the transposed matrix.

5. The method for collecting core information based on close-range photography according to claim 1, characterized in that: The specific process of step S5 is: based on the collinearity condition equation, the image exterior orientation elements and the three-dimensional coordinates of the object point are jointly solved through the image point observation values ​​on multiple images, and the image exterior parameters are solved and the sparse point cloud of the object point is formed through the least squares adjustment and combined control points.

6. The method for collecting core information based on close-range photography according to claim 1, characterized in that: The specific process of step S6 is as follows: generate the depth value of each pixel through pixel-level matching of multi-view images, optimize the depth map based on semi-global matching, convert each pixel into the coordinates in the camera coordinate system in combination with the camera intrinsic parameters, and project it to the global coordinate system through the camera extrinsic parameters, and finally improve the point cloud quality through filtering, fusion, and isolation post-processing steps to fuse into a dense point cloud.

7. The method for collecting core information based on close-range photography according to claim 1, characterized in that: The specific process of step S7 is as follows: by deleting outliers whose distance to the mean exceeds 3σ, isolated points or duplicate points with insufficient points in the field are removed, and the dense point cloud is further denoised, filtered and resampled; then the point cloud normal vector is calculated based on the KD-Tree neighborhood, an octree structure is constructed, and the point cloud is converted into an indicator function gradient field; and the Poisson equation is solved to extract the isosurface to generate a three-dimensional grid of the core.

8. The method for collecting core information based on close-range photography according to claim 1, characterized in that: The step S8 comprises the following steps: S81, determining visibility of each mesh face or vertex based on the depth map; S82, use bilinear interpolation and multi-view color fusion to sample colors from visible images and fuse them into 3D mesh surfaces and point clouds; S83. Project three identical textures or different textures onto the model surface along three axes; perform weighted blending on the three projection results according to the surface normal direction to ensure a natural texture transition; and perform final color calculation to balance the brightness and color difference of different images, and finally generate the core surface texture.

9. The method for collecting core information based on close-range photography according to claim 8, characterized in that: The calculation formula in step S83 includes: Where: P x , P y , P z is the two-dimensional coordinate of the three-dimensional point on the projection plane of each axis; t is the number of texture repetitions; o is the texture translation offset; w x , w y , w z are the weight coefficients of the three axes; n is the surface normal vector; X axis , Y axis , Z axis are the unit vectors of the three axes; k is the sharpness factor; T x , T y , T z The texture sampling function for the three axes.

10. The method for collecting core information based on close-range photography according to claim 1, characterized in that: The specific process of step S9 is as follows: for the three-dimensional grid and color texture point cloud of the core, when the core cylinder is cut along a vertical line and unfolded into a plane rectangle, each point on the core surface is mapped to a point on the plane to form a core unfolded image.

Images