Columnar core outer surface hyperspectral image reconstruction method

Abstract

The invention belongs to the technical field of rock core scanning and earth science, and particularly relates to a columnar rock core outer surface hyperspectral image reconstruction method, which comprises the following steps of: 1, acquiring a columnar rock core multi-angle hyperspectral image; 2, cutting an effective area of the multi-angle hyperspectral image of the columnar core; step 3, carrying out radiation correction on the multi-angle hyperspectral image of the columnar rock core; step 4, geometric correction of the multi-angle hyperspectral image of the columnar rock core; 5, splicing the multi-angle hyperspectral images of the columnar rock core; and step 6, removing an overlapping region of the spliced image to obtain a cylindrical core outer surface hyperspectral image reconstruction result. According to the method, radiation brightness distortion and geometric shape distortion generated in the hyperspectral image scanning process can be effectively corrected, and therefore accurate reconstruction of the hyperspectral image on the outer surface of the whole columnar rock core is achieved.

Description

Technical Field

[0001] The invention belongs to the technical field of core scanning and earth science, and in particular relates to a method for reconstructing a hyperspectral image of the outer surface of a columnar core. Background Art

[0002] Drill cores are the most important and intuitive physical geological data for studying the structure of underground rock layers and understanding underground mineral conditions. However, the cores are easily weathered and damaged after being taken out of the barrel, which reduces the value of the cores for continued research and use. Core hyperspectral scanning is an effective way to manage and store cores. It can not only fully reproduce the appearance characteristics of the cores and achieve permanent preservation, but also perform spectral analysis on the cores, which is more conducive to rock and mineral identification.

[0003] There are two main ways to scan the outer surface image of the core: the first is that the core is stationary, and the sensor performs 360° scanning and imaging around the core; the second is that the sensor is fixed, the core is flipped for scanning and imaging at multiple angles, and then several images are stitched together to complete the reconstruction of the outer surface image of the entire columnar core. Most existing methods use CCD cameras to take red, green, and blue three-channel images. The camera has a large field of view, fast imaging speed, and relatively small data volume. Therefore, the first scanning method is mainly used. However, for hyperspectral sensors with a small field of view and slow imaging speed, they have hundreds of spectral bands and a large amount of data, which are not suitable for the first scanning method. When the second scanning method of flipping the core is used, the image often produces obvious radiation distortion and geometric distortion, and there is a lack of effective correction methods to help restore the hyperspectral image of the outer surface of the columnar core.

[0004] Therefore, it is urgent to develop a method for reconstructing the hyperspectral image of the outer surface of the columnar core, which can effectively correct the radiation brightness distortion and geometric shape distortion generated during the hyperspectral image scanning process, so as to achieve accurate reconstruction of the hyperspectral image of the entire columnar core outer surface. Summary of the invention

[0005] The purpose of the present invention is to provide a method for reconstructing a hyperspectral image of the outer surface of a columnar core, which can effectively correct the radiation brightness distortion and geometric shape distortion generated during the hyperspectral image scanning process, thereby realizing accurate reconstruction of the hyperspectral image of the entire columnar core outer surface.

[0006] The technical solution to achieve the purpose of the present invention is:

[0007] A method for reconstructing a hyperspectral image of the outer surface of a columnar core, the method comprising:

[0008] Step 1: Multi-angle hyperspectral image acquisition of columnar cores;

[0009] Step 2: cropping of effective area of ​​multi-angle hyperspectral image of columnar core;

[0010] Step 3: Radiometric correction of multi-angle hyperspectral images of columnar cores;

[0011] Step 4: Geometric correction of multi-angle hyperspectral images of columnar cores;

[0012] Step 5: stitching of multi-angle hyperspectral images of columnar cores;

[0013] Step 6: Remove the overlapping areas of the stitched images to obtain the hyperspectral image reconstruction result of the outer surface of the columnar core.

[0014] The step 1 specifically includes: rotating the core at a fixed angle, scanning and imaging the core multiple times, collecting high-spectral images of the columnar core at multiple angles, and obtaining multiple core-angle cylindrical high-spectral images A.

[0015] The angle of core rotation in step 1 is no greater than 90°.

[0016] The step 2 comprises:

[0017] Step 2.1: Use software to count the starting row, ending row, starting column, and ending column of each core angle cylindrical hyperspectral image A;

[0018] Step 2.2: Obtain the range data from the starting row to the ending row and the range data from the starting column to the ending column in each core angle cylindrical hyperspectral image A, that is, the effective area image B of each core angle cylindrical hyperspectral image A.

[0019] The step 3 comprises:

[0020] Step 3.1: Make a diffuse reflection rod with the same length and diameter as the core in step 1, and obtain the diffuse reflection rod angle 1 cylindrical hyperspectral image Z1 in the same way as the core measurement in step 1.

[0021] Step 3.2: Calculate the radiation correction scale factor The calculation formula is

[0022]

[0023] Where f(·) represents the average value of all pixels in the center column of the diffuse reflection stick image Z1 located directly below the sensor, and g(e) represents the average value of all pixels in the e-th column of the diffuse reflection stick image Z1.

[0024] Step 3.3: Perform radiation correction on the effective area image B of each core angle cylindrical hyperspectral image using the following formula to obtain the radiation corrected cylindrical hyperspectral image C of each core angle:

[0025]

[0026] Where h(e)' represents the pixel value of the core angle cylindrical hyperspectral image after radiation correction, and h(e) represents the pixel value of the effective area image of the core angle cylindrical hyperspectral image.

[0027] The step 4 comprises:

[0028] Step 4.1: With the tangent direction of the core upper surface as the X-axis and the perpendicular direction of the core upper surface as the Y-axis, a plane rectangular coordinate system is established. The hyperspectral sensor and light source are located on the Y-axis. The coordinates (x, y) of any point P on the outer surface of the core cross section are calculated;

[0029] The coordinates (x, y) of point P are calculated as follows:

[0030]

[0031]

[0032] Wherein, u represents the distance between the original position of the image of point P on the hyperspectral sensor and the origin O of the plane rectangular coordinate system; h represents the distance between the origin O of the plane rectangular coordinate system and the hyperspectral sensor and light source; r represents the radius of the columnar core;

[0033] Step 4.2: According to the coordinates (x, y) of any point P on the outer surface of the core cross section, the original position of the point P on the radiation-corrected cylindrical core multi-angle hyperspectral image C is geometrically corrected to obtain the geometrically corrected position of the point P on the radiation-corrected cylindrical core multi-angle hyperspectral image C;

[0034] Calculate the distance between any point P on the outer surface of the core cross section and the origin O of the plane rectangular coordinate system, that is, the length of the line segment OP: the length of the line segment OP |OP| is

[0035]

[0036] According to the length of line segment OP, calculate the length of arc l corresponding to line segment OP on the circumference of the outer surface of the core cross section: the central angle α corresponding to line segment OP is

[0037]

[0038] The length of arc l corresponding to line segment OP is

[0039] l = α·r,

[0040] The length of arc l is the distance between the geometrically corrected position of any point P on the hyperspectral image and the origin O, and the length of arc l is the geometrically corrected position of point P on the hyperspectral image after radiation correction;

[0041] Step 4.3: According to steps 4.1-4.2, geometric correction is performed on each pixel point on the radiation-corrected columnar core multi-angle hyperspectral image C obtained in step 3.3 to obtain the geometrically corrected position of each pixel point, and the geometrically corrected image D is obtained according to the geometrically corrected position of each pixel point;

[0042] Step 4.4: Use the linear interpolation method to calculate the pixel value of each point in the image D after each geometric position correction, and obtain the image E after each geometric position correction and pixel assignment. The calculation formula is:

[0043]

[0044] Where Y i represents the pixel value of point i in image E after geometric position correction and pixel assignment, y i-1 Represents the pixel value of point i-1 in the same row as point i in image C before geometric position correction, y i+1 represents the pixel value of point i+1 in the same row as point i in image C before geometric position correction, x i represents the horizontal coordinate position of point i in the row of image D after geometric position correction, x i-1 represents the horizontal coordinate position of the row where point i-1 is located in the image D after geometric position correction, x i+1 The horizontal coordinate position of the row where the representative point i+1 is located in the image D after geometric position correction.

[0045] Step 4.5: Perform effective area cropping on each image E after geometric position correction and pixel assignment, and obtain the cropped columnar core multi-angle hyperspectral image F.

[0046] The step 4.5 comprises:

[0047] Step 4.5.1: Referring to step 1, rotate clockwise for a fixed angle each time to collect n+1 sets of image data for the columnar core, which are recorded as U0, U1, U2, ..., U n , n is an integer not less than 3;

[0048] Step 4.5.2: Repeat steps 4.1 to 4.4, and transform images U0, U1, U2, ..., U n Perform geometric correction to obtain geometrically corrected images V0, V1, V2, ..., V n ;

[0049] Step 4.5.3: Let the center column of image V0 be Where m is the column number of the center column of image V0; In images V1, V2, ..., V n The columns with the same name in Among them, S1, S2, ..., S n is the column number of the column with the same name in the corresponding image;

[0050] Step 4.5.4: Calculate separately and The similarity of n ;

[0051] Step 4.5.5: Set the threshold δ, when the similarity φ n+1 is less than the threshold δ, and the similarity φ n When it is greater than or equal to the threshold δ, the similarity φ is determined n Corresponding column data After geometric correction, no information loss occurs and it is considered valid data, that is, the end column number of the valid area is S n ;

[0052] Step 4.5.6: Calculate the starting column number of the effective area by the symmetry principle as 2m-S n , so the number of columns of valid data ranges from List;

[0053] Step 4.5.7: For each geometric position corrected and pixel assigned image E, The column data is cropped to obtain the multi-angle hyperspectral image F of the columnar core after the effective area is cropped.

[0054] In step 4.5.1, the angle of each clockwise rotation of the core is 10°.

[0055] The step 5 comprises:

[0056] Step 5.1: Count the position information of the same-name points in two adjacent columnar core images;

[0057] Step 5.2: Align the columns of the positions of the same-name points in the two adjacent columnar core images, and use the overlapping area pixel value calculation formula to smooth the overlapping area of ​​the adjacent columnar core images, and complete the splicing of the columnar core multi-angle hyperspectral image F after the effective area is cut, and the spliced ​​image is recorded as G;

[0058] The calculation formula of pixel value in overlapping area is:

[0059]

[0060] Where a is the starting column horizontal coordinate of the overlapping area, b is the ending column horizontal coordinate of the overlapping area, f(c, d) and g(c, d) represent the pixel values ​​at the dth row and cth column of the two adjacent columnar core images, and p(c, d) represents the pixel value at the dth row and cth column in the overlapping area.

[0061] The specific step 6 is: respectively delete the left and right edges of the spliced ​​image G The data is then recorded to obtain the final hyperspectral image reconstruction result of the outer surface of the columnar core, which is recorded as image H.

[0062] The beneficial technical effects of the present invention are:

[0063] 1. The present invention provides a method for reconstructing a hyperspectral image of the outer surface of a columnar core, which can effectively correct the radiation brightness distortion and geometric shape distortion generated during the hyperspectral image scanning process, thereby realizing accurate reconstruction of the hyperspectral image of the outer surface of the entire columnar core.

[0064] 2. The present invention provides a method for reconstructing a hyperspectral image of the outer surface of a columnar core, which can eliminate background area interference by effectively cropping the collected multi-angle hyperspectral image of the columnar core.

[0065] 3. The present invention provides a method for reconstructing a hyperspectral image of the outer surface of a columnar core. By performing radiation correction on the multi-angle hyperspectral image of the columnar core, the brightness difference between the edge and the center of the image can be significantly improved, making the overall brightness of the image more uniform and closer to the actual situation of the outer surface of the core, and presenting a more realistic core appearance image to the staff.

[0066] 4. The present invention provides a method for reconstructing a hyperspectral image of the outer surface of a columnar core. By geometrically correcting the multi-angle hyperspectral image of the columnar core, the geometric shape distortion of the columnar core is effectively reduced or avoided. By effectively cropping the image after geometric correction and pixel assignment, the information loss caused by image compression is reduced, thereby ensuring that the columnar core image is clear.

[0067] 5. The present invention provides a method for reconstructing a hyperspectral image of the outer surface of a columnar core. By splicing the hyperspectral image after geometric correction and effective area cropping, the splicing marks in the overlapping area are reduced, there is no significant brightness difference, and the image is smoother. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] Figure 1 A flow chart of a method for reconstructing a hyperspectral image of the outer surface of a columnar core provided by the present invention;

[0069] Figure 2 Schematic diagram of the process of the method for reconstructing the outer surface hyperspectral image of a columnar core in an embodiment of the present invention;

[0070] Figure 3 Schematic diagram of the geometric coordinates of a columnar core in an embodiment of the present invention;

[0071] Figure 4Schematic diagram of the splicing method of the hyperspectral image of the outer surface of the columnar core in an embodiment of the present invention. DETAILED DESCRIPTION

[0072] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.

[0073] In a method for reconstructing a hyperspectral image of the outer surface of a columnar core provided by the present invention, since the surface of the core is a columnar arc surface, if its 360° complete unfolding surface is to be obtained, it is necessary to splice the scanned images, which requires that the images must have overlapping parts. Therefore, it is necessary to rotate the core 120° at least three times to perform 360° scanning and imaging of the core. However, when the core is rotated 120° and scanned three times, after the invalid parts with serious information loss on both sides of the edge of the image are cut off, there is no overlapping part between the three scanning results, and it is impossible to achieve image splicing to obtain a 360° complete unfolding surface. Therefore, to achieve image splicing to obtain a 360° complete unfolding surface, it is necessary to scan at least four times and obtain four images for splicing, so as to meet the requirement that after removing the invalid edge area, there is an overlapping area for splicing between two adjacent images. Therefore, rotating the core 90° and scanning the core 360° four times, rotating the core 60° and scanning the core 360° six times, or even rotating the core 30° and scanning the core 360° 12 times can also meet the splicing requirements.

[0074] The method of the present invention is specifically described below by taking the example of rotating the core by 90° and scanning the core 360° four times.

[0075] like Figure 1 As shown, the present invention provides a method for reconstructing a hyperspectral image of the outer surface of a columnar core, which specifically includes the following methods:

[0076] Step 1: Multi-angle hyperspectral image acquisition of columnar cores.

[0077] The core is rotated at a fixed angle, and the core is scanned and imaged multiple times to collect high-spectral images of the columnar core at multiple angles. Figure 2 As shown, in this embodiment, the core is scanned 360° four times by rotating the core 90° to collect hyperspectral images of a total of four cylinders at four angles of the core, which are respectively recorded as core angle 1 cylindrical hyperspectral image A1, core angle 2 cylindrical hyperspectral image A2, core angle 3 cylindrical hyperspectral image A3, and core angle 4 cylindrical hyperspectral image A4.

[0078] Step 2: Cropping of effective areas of multi-angle hyperspectral images of columnar cores.

[0079] The effective area of ​​the multi-angle hyperspectral image of the columnar core collected in step 1 is automatically cropped to eliminate the interference of the background area. Since the measured core diameter is consistent and the position of the hyperspectral sensor is relatively fixed, the position of the core in the multi-angle cylindrical image collected by the rotating core is fixed. We only need to manually count the number of rows and columns where the edge of the effective area of ​​the core is located in the image in the software to achieve automatic cropping of the effective area of ​​all columnar core multi-angle hyperspectral images.

[0080] The details are as follows:

[0081] Step 2.1: In the ENVI software, the starting row of the core angle 1 cylindrical hyperspectral image A1 is counted as i1, the ending row is i2, the starting column is j1, and the ending column is j2.

[0082] Step 2.2: Obtain the data in the range of rows i1 to i2 and columns j1 to j2 in image A1, which is the effective area image B1 of the core angle 1 cylindrical hyperspectral image A1.

[0083] Step 2.3: Using the same range of rows and columns as step 2.1, automatically crop the effective area image B2 of the cylindrical hyperspectral image A2 of core angle 2, the effective area image B3 of the cylindrical hyperspectral image A3 of core angle 3, and the effective area image B4 of the cylindrical hyperspectral image A4 of core angle 4.

[0084] Step 3: Radiometric correction of multi-angle hyperspectral images of cylindrical cores.

[0085] Step 3.1: Make a diffuse reflection rod with the same length and diameter as the core in step 1, and obtain the diffuse reflection rod angle 1 cylindrical hyperspectral image Z1 in the same way as the core measurement in step 1.

[0086] Step 3.2: Calculate the radiation correction scale factor The calculation formula is

[0087]

[0088] Where f(·) represents the average value of all pixels in the center column of the diffuse reflection stick image Z1 located directly below the sensor, and g(e) represents the average value of all pixels in the e-th column of the diffuse reflection stick image Z1.

[0089] Step 3.3: Perform radiation correction on the core angle 1 cylindrical effective area image B1, the core angle 2 cylindrical effective area image B2, the core angle 3 cylindrical effective area image B3 and the core angle 4 cylindrical effective area image B4 using the following formula to obtain the radiation-corrected core angle 1 cylindrical hyperspectral image C1, the core angle 2 cylindrical hyperspectral image C2, the core angle 3 cylindrical hyperspectral image C3 and the core angle 4 cylindrical hyperspectral image C4:

[0090]

[0091] Where h(e)' represents the pixel value of the core angle cylindrical hyperspectral image after radiation correction, and h(e) represents the pixel value of the effective area image of the core angle cylindrical hyperspectral image.

[0092] Step 4: Geometric correction of multi-angle hyperspectral images of cylindrical cores.

[0093] Step 4.1: With the tangent direction of the core upper surface as the X-axis and the perpendicular direction of the core upper surface as the Y-axis, a plane rectangular coordinate system is established. The hyperspectral sensor and light source are located on the Y-axis. The coordinates (x, y) of any point P on the outer surface of the core cross section are calculated;

[0094] like Figure 3 As shown in the figure, with the tangent direction of the upper surface of the core as the X-axis and the perpendicular direction of the upper surface of the core as the Y-axis, a plane rectangular coordinate system is established, and its origin is recorded as point O; the hyperspectral sensor and light source are located on the Y-axis, recorded as point T; the center point of the core cross section is recorded as point O'; P is a point on the outer surface of the columnar core cross section, and its coordinates are recorded as (x, y); M is the original position of the image formed by point P on the hyperspectral sensor, that is, M is the intersection of the straight line TP and the X-axis; the distance between the original position of the image formed by point P on the hyperspectral sensor and the origin of the coordinate axis is recorded as u, that is, the distance between point M and the origin O is recorded as u; the distance between the origin of the coordinate axis and the hyperspectral sensor and light source is recorded as h, that is, the distance between point O and point T is recorded as h; the radius of the columnar core is recorded as r.

[0095] In the plane rectangular coordinate system, the equation of the line TP is

[0096]

[0097] The equation of the cross-section circle of a cylindrical core is

[0098] x 2 +(y+r) 2 =r 2 (4)

[0099] The coordinates of point P are obtained by formula (3) and formula (4):

[0100]

[0101]

[0102] Step 4.2: According to the coordinates (x, y) of any point P on the outer surface of the core cross section, the original position of the point P on the radiation-corrected cylindrical core multi-angle hyperspectral image C1, C2, C3 or C4 is geometrically corrected to obtain the geometrically corrected position of the point P on the radiation-corrected cylindrical core multi-angle hyperspectral image C1, C2, C3 or C4;

[0103] like Figure 3 As shown, N is the corrected position of the hyperspectral image after point M is geometrically corrected, that is, point N is the position of the hyperspectral image on the X-axis after geometric correction.

[0104] The original position of point P on the multi-angle hyperspectral image C1, C2, C3 or C4 of the columnar core after radiation correction, that is, the intersection M of the straight line TP and the X-axis, is geometrically corrected by stretching point M to obtain the geometrically corrected point N.

[0105] The specific process of geometrically correcting point M to obtain the geometrically corrected point N is as follows: calculate the distance between any point P on the outer surface of the core cross section and the origin O of the plane rectangular coordinate system, that is, the length of the line segment OP; calculate the length of the arc l corresponding to the line segment OP on the circumference of the outer surface of the core cross section according to the length of the line segment OP; the length of the arc l is the distance between the geometrically corrected position N of any point P on the hyperspectral image and the origin O; the length value of the arc l is the geometrically corrected position of point N on the X-axis (the geometrically corrected position of point P on the hyperspectral image after radiation correction).

[0106] The length of line segment OP |OP| is

[0107]

[0108] The central angle α corresponding to the line segment OP is

[0109]

[0110] The length of arc l corresponding to line segment OP is

[0111] l=α·r (9)

[0112] The length of line segment ON is equal to the length of arc l, that is, |ON|=l.

[0113] The l value is the geometrically corrected position of point P in the radiometrically corrected hyperspectral image.

[0114] Step 4.3: According to steps 4.1-4.2, geometric correction is performed on each pixel point on the radiation-corrected columnar core multi-angle hyperspectral images C1, C2, C3, and C4 obtained in step 3.3 to obtain the geometrically corrected position of each pixel point, and geometrically corrected images D1, D2, D3, and D4 are obtained according to the geometrically corrected position of each pixel point;

[0115] Step 4.4: Use the linear interpolation method to calculate the pixel value of each point in the images D1, D2, D3, and D4, and obtain the images E1, E2, E3, and E4 after geometric position correction and pixel assignment. The calculation formula is:

[0116]

[0117] Where Y i represents the pixel value of point i in images E1, E2, E3, and E4 after geometric position correction and pixel assignment, y i-1 represents the pixel value of point i-1 in the same row as point i in images C1, C2, C3, and C4 before geometric position correction, y i+1 represents the pixel value of point i+1 in the same row as point i in images C1, C2, C3, and C4 before geometric position correction, x i represents the horizontal coordinate position of the row where point i is located in the images D1, D2, D3, and D4 after geometric position correction, x i-1 The horizontal coordinate position of the row where the point i-1 is located in the images D1, D2, D3, and D4 after geometric position correction, x i+1 The horizontal coordinate position of the row where the representative point i+1 is located in the images D1, D2, D3, and D4 after geometric position correction.

[0118] Step 4.5: Perform effective area cropping on images E1, E2, E3, and E4 respectively to obtain cropped columnar core multi-angle hyperspectral images F1, F2, F3, and F4.

[0119] The images after geometric position correction and pixel assignment are not completely effective. The closer to the edges of the columnar core, the blurrier the image and the more serious the information loss. The lost data cannot be restored simply through geometric correction, and the information loss can only be compensated by multiple multi-angle shooting. Therefore, the images of the edges of the core can only be discarded, and only the effective area is cut for subsequent processing. Specific effective area selection method:

[0120] In order to compensate for information loss through multiple multi-angle shooting, multiple groups of multi-angle images are collected to verify the validity of the edge data on both sides of the core. Considering that the smaller the angle of single rotation of the core, the higher the accuracy of the determined effective area, the rotation angle is set to 10° in this embodiment.

[0121] Step 4.5.1: Referring to step 1, rotate 10° clockwise each time to collect 10 sets of image data for the columnar core, which are recorded as U0, U1, U2, U3, U4, U5, U6, U7, U8, and U9 respectively.

[0122] Step 4.5.2: Repeat steps 4.1 to 4.4, and perform geometric correction on images U0, U1, U2, U3, U4, U5, U6, U7, U8, and U9, respectively, to obtain geometrically corrected images V0, V1, V2, V3, V4, V5, V6, V7, V8, and V9.

[0123] Step 4.5.3: Let the center column of image V0 be Where m is the column number of the center column of image V0 (m is an integer greater than 0, for example, image V0 has 121 columns, and the value of m is 61). The columns with the same name in images V1, V2, V3, V4, V5, V6, V7, V8, and V9 are denoted as Among them, S1, S2, S3, S4, S5, S6, S7, S8, and S9 are the column numbers of the columns with the same name in the corresponding images.

[0124] Step 4.5.4: Calculate separately and The similarity of n (n=1,2,…9).

[0125] Step 4.5.5: Set the threshold δ, when the similarity φ n+1 is less than the threshold δ, and the similarity φ n When it is greater than or equal to the threshold δ, the similarity φ is determined n Corresponding column data After geometric correction, no information loss occurs and it is considered valid data, that is, the end column number of the valid area is S n .

[0126] Step 4.5.6: Calculate the starting column number of the effective area by the symmetry principle as 2m-S n , so the number of columns of valid data ranges from List.

[0127] Step 4.5.7: For images E1, E2, E3, and E4, The column data are cropped to obtain multi-angle hyperspectral images F1, F2, F3, and F4 of the columnar core after effective area cropping.

[0128] Step 5: Multi-angle hyperspectral image stitching of columnar cores.

[0129] Step 5.1: Count the position information of the same-name points in two adjacent columnar core images. Since the core is rotated at a fixed angle (i.e., 90°) each time in step 1, the positions of the same-name points in two adjacent images in F1, F2, F3, and F4 are fixed. Specifically, the same-name point 1 is located in the m1th column of the core image F1 and in the m2th column of the core image F2; the same-name point 2 is located in the m3th column of the core image F2 and in the m4th column of the core image F3; the same-name point 3 is located in the m5th column of the core image F3 and in the m6th column of the core image F4.

[0130] Step 5.2: Align the columns of the positions of the same-name points in the two adjacent columnar core images, and use formula (11) to smooth the overlapping areas of the adjacent columnar core images to complete the splicing of the columnar core multi-angle hyperspectral image F after the effective area is cropped. The spliced ​​image is recorded as G.

[0131] Step 5.2.1: Align the m1th column of the core image F1 and the m2th column of the core image F2, and use formula (11) to smooth the overlapping area of ​​the image F1 and the image F2.

[0132] Step 5.2.2: Align the m3th column of the core image F2 and the m4th column of the core image F3, and use formula (11) to smooth the overlapping area of ​​the image F2 and the image F3.

[0133] Step 5.2.3: Align the m5th column of the core image F3 and the m6th column of the core image F4, and use formula (11) to smooth the overlapping area of ​​the image F3 and the image F4.

[0134] At this point, the splicing of the four cylindrical hyperspectral images F1, F2, F3, and F4 at four angles of the columnar core is completed, and the spliced ​​image is denoted as G.

[0135] In order to alleviate the stitching marks that appear during the stitching process, the image is smoothed by performing weighted average processing on the image. Figure 4 Assume that the horizontal coordinate of the starting column of the overlapping area is a, and the horizontal coordinate of the ending column is b, the calculation formula for the pixel value at the dth row and cth column in the overlapping area is

[0136]

[0137] Where f(c, d) and g(c, d) represent the pixel values ​​at the dth row and cth column of two adjacent columnar core images (e.g., core 1 and core 2), and p(c, d) represents the pixel value at the dth row and cth column in the overlapping area.

[0138] Step 6: Remove the overlapping areas of the stitched images to obtain the hyperspectral image reconstruction result of the outer surface of the columnar core.

[0139] There are still some overlapping areas on the left and right ends of the stitched image G. In order to obtain an accurate 360° unfolded image of the outer surface of the columnar core, it is necessary to delete the left and right edges of the stitched image G respectively. The data is then recorded to obtain the final hyperspectral image reconstruction result of the outer surface of the columnar core, which is recorded as image H.

[0140] The present invention is described in detail above with reference to the accompanying drawings and embodiments, but the present invention is not limited to the above embodiments, and various changes can be made within the knowledge of ordinary technicians in the field without departing from the purpose of the present invention. The contents not described in detail in the present invention can adopt the existing technology.

Claims

1. A method for reconstructing a hyperspectral image of the outer surface of a columnar core, characterized in that: The method comprises: Step 1: Multi-angle hyperspectral image acquisition of columnar cores; Step 2: cropping of effective area of ​​multi-angle hyperspectral image of columnar core; Step 3: Radiometric correction of multi-angle hyperspectral images of columnar cores; Step 4: Geometric correction of multi-angle hyperspectral images of columnar cores; Step 5: stitching of multi-angle hyperspectral images of columnar cores; Step 6: Remove the overlapping areas of the stitched images to obtain the hyperspectral image reconstruction result of the outer surface of the columnar core.

2. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 1, characterized in that: The step 1 specifically includes: rotating the core at a fixed angle, scanning and imaging the core multiple times, collecting high-spectral images of the columnar core at multiple angles, and obtaining multiple core-angle cylindrical high-spectral images A.

3. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 1, characterized in that: The angle of core rotation in step 1 is no greater than 90°.

4. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 1, characterized in that: The step 2 comprises: Step 2.1: Use software to count the starting row, ending row, starting column, and ending column of each core angle cylindrical hyperspectral image A; Step 2.2: Obtain the range data from the starting row to the ending row and the range data from the starting column to the ending column in each core angle cylindrical hyperspectral image A, that is, the effective area image B of each core angle cylindrical hyperspectral image A.

5. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 1, characterized in that: The step 3 comprises: Step 3.1: Make a diffuse reflection rod with the same length and diameter as the core in step 1, and obtain the diffuse reflection rod angle 1 cylindrical hyperspectral image Z1 in the same way as the core measurement in step 1. Step 3.2: Calculate the radiation correction scale factor The calculation formula is Where f(·) represents the average value of all pixels in the center column of the diffuse reflection stick image Z1 located directly below the sensor, and g(e) represents the average value of all pixels in the e-th column of the diffuse reflection stick image Z1. Step 3.3: Perform radiation correction on the effective area image B of each core angle cylindrical hyperspectral image using the following formula to obtain the radiation corrected cylindrical hyperspectral image C of each core angle: Where h(e)' represents the pixel value of the core angle cylindrical hyperspectral image after radiation correction, and h(e) represents the pixel value of the effective area image of the core angle cylindrical hyperspectral image.

6. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 1, characterized in that: The step 4 comprises: Step 4.1: With the tangent direction of the core upper surface as the X-axis and the perpendicular direction of the core upper surface as the Y-axis, a plane rectangular coordinate system is established. The hyperspectral sensor and light source are located on the Y-axis. The coordinates (x, y) of any point P on the outer surface of the core cross section are calculated; The coordinates (x, y) of point P are calculated as follows: Wherein, u represents the distance between the original position of the image of point P on the hyperspectral sensor and the origin O of the plane rectangular coordinate system; h represents the distance between the origin O of the plane rectangular coordinate system and the hyperspectral sensor and light source; r represents the radius of the columnar core; Step 4.2: According to the coordinates (x, y) of any point P on the outer surface of the core cross section, the original position of the point P on the radiation-corrected cylindrical core multi-angle hyperspectral image C is geometrically corrected to obtain the geometrically corrected position of the point P on the radiation-corrected cylindrical core multi-angle hyperspectral image C; Calculate the distance between any point P on the outer surface of the core cross section and the origin O of the plane rectangular coordinate system, that is, the length of the line segment OP: the length of the line segment OP |OP| is According to the length of line segment OP, calculate the length of arc l corresponding to line segment OP on the circumference of the outer surface of the core cross section: the central angle α corresponding to line segment OP is The length of arc l corresponding to line segment OP is l＝α·r， The length of arc l is the distance between the geometrically corrected position of any point P on the hyperspectral image and the origin O, and the length of arc l is the geometrically corrected position of point P on the hyperspectral image after radiation correction; Step 4.3: According to steps 4.1-4.2, geometric correction is performed on each pixel point on the radiation-corrected columnar core multi-angle hyperspectral image C obtained in step 3.3 to obtain the geometrically corrected position of each pixel point, and the geometrically corrected image D is obtained according to the geometrically corrected position of each pixel point; Step 4.4: Use the linear interpolation method to calculate the pixel value of each point in the image D after each geometric position correction, and obtain the image E after each geometric position correction and pixel assignment. The calculation formula is: Where Y i represents the pixel value of point i in image E after geometric position correction and pixel assignment, y i-1 Represents the pixel value of point i-1 in the same row as point i in image C before geometric position correction, y i+1 represents the pixel value of point i+1 in the same row as point i in image C before geometric position correction, x i represents the horizontal coordinate position of point i in the row of image D after geometric position correction, x i-1 represents the horizontal coordinate position of the row where point i-1 is located in the image D after geometric position correction, x i+1 The horizontal coordinate position of the row where the representative point i+1 is located in the image D after geometric position correction. Step 4.5: Perform effective area cropping on each image E after geometric position correction and pixel assignment, and obtain the cropped columnar core multi-angle hyperspectral image F.

7. A method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 6, characterized in that: The step 4.5 comprises: Step 4.5.1: Referring to step 1, rotate clockwise for a fixed angle each time to collect n+1 sets of image data for the columnar core, which are recorded as U0, U1, U2, ..., U n , n is an integer not less than 3; Step 4.5.2: Repeat steps 4.1 to 4.4, and transform images U0, U1, U2, ..., U n Perform geometric correction to obtain geometrically corrected images V0, V1, V2, ..., V n ; Step 4.5.3: Let the center column of image V0 be Where m is the column number of the center column of image V0; In images V1, V2, ..., V n The columns with the same name in Among them, S1, S2, ..., S n is the column number of the column with the same name in the corresponding image; Step 4.5.4: Calculate separately and The similarity of n ; Step 4.5.5: Set the threshold δ, when the similarity φ n+1 is less than the threshold δ, and the similarity φ n When it is greater than or equal to the threshold δ, the similarity φ is determined n Corresponding column data After geometric correction, no information loss occurs and it is considered valid data, that is, the end column number of the valid area is S n ; Step 4.5.6: Calculate the starting column number of the effective area by the symmetry principle as 2m-S n , so the number of columns of valid data ranges from List; Step 4.5.7: For each geometric position corrected and pixel assigned image E, The column data is cropped to obtain the multi-angle hyperspectral image F of the columnar core after the effective area is cropped.

8. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 7, characterized in that: In step 4.5.1, the angle of each clockwise rotation of the core is 10°.

9. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 1, characterized in that: The step 5 comprises: Step 5.1: Count the position information of the same-name points in two adjacent columnar core images; Step 5.2: Align the columns of the positions of the same-name points in the two adjacent columnar core images, and use the overlapping area pixel value calculation formula to smooth the overlapping area of ​​the adjacent columnar core images, and complete the splicing of the columnar core multi-angle hyperspectral image F after the effective area is cut, and the spliced ​​image is recorded as G; The calculation formula of pixel value in overlapping area is: Where a is the starting column horizontal coordinate of the overlapping area, b is the ending column horizontal coordinate of the overlapping area, f(c, d) and g(c, d) represent the pixel values ​​at the dth row and cth column of the two adjacent columnar core images, and p(c, d) represents the pixel value at the dth row and cth column in the overlapping area.

10. The method for reconstructing a hyperspectral image of the outer surface of a columnar core according to claim 9, characterized in that: The specific step 6 is: respectively delete the left and right edges of the spliced ​​image G The data is then recorded to obtain the final hyperspectral image reconstruction result of the outer surface of the columnar core, which is recorded as image H.

Images