Core image color correction and profile construction method and system

Abstract

The present application relates to the technical field of image processing, and discloses a core image color correction and profile construction method and system, comprising: extracting an intrinsic reflectance feature map of a visible light sequence image stripped of illumination geometric shadow and highlight; extracting reflectance-geometry consistency feature points of a core to construct a robust feature point set of the core; constructing a joint energy function of the visible light sequence image, combining the intrinsic reflectance feature map and the robust feature point set, synchronously solving a color correction parameter matrix and a geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence doubly normalized in color and geometry; in an overlapping area of the intermediate image sequence, calculating an adaptive seam line according to the intrinsic reflectance feature map; using the adaptive seam line to perform reflectance confidence weighted fusion on high-frequency components and low-frequency components of the decomposed intermediate image sequence to reconstruct a core digital profile image of the core. The present application can improve the accuracy of core image color correction and profile.

Description

Technical Field

[0001] This invention relates to a method and system for color correction and profile construction of rock core images, belonging to the field of image processing technology. Background Technology

[0002] Core image color correction and profile construction refers to the process of using advanced computer vision and image processing technologies to perform illumination removal, color restoration, and distortion correction on acquired core sequence images, and then precisely stitching and fusing them to generate high-fidelity, continuous, and visualized two-dimensional or three-dimensional digital core profiles. This method can not only overcome color distortion and geometric misalignment caused by the field shooting environment and the physical characteristics of the core surface, achieving permanent digital storage and lossless sharing of core geological information, but also assist geological experts in accurately identifying lithological bedding, structural fractures, and mineral composition through standardized digital profiles, thereby significantly improving the efficiency and accuracy of oil and gas exploration, mineral resource development, and geological scientific research.

[0003] Traditional core image color correction and profile construction typically employ manual photography or single-light source linear array scanning. Color correction is performed using histogram-equalized color chart matching, and image stitching is done using texture feature-based algorithms such as SIFT and SURF. This approach is susceptible to interference from non-Lambertian reflections on the core surface, uneven lighting, and shadows, resulting in "false colors" even after color correction. Furthermore, when dealing with homogeneous sandstone and other areas with simple textures, traditional feature matching algorithms often fail due to a lack of sufficient feature points, causing stitching misalignment or stratigraphic fractures. Consequently, the color correction and profile accuracy of digital core profiles are poor. Summary of the Invention

[0004] This invention provides a method and system for color correction and profile construction of rock core images, the main purpose of which is to improve the accuracy of color correction and profile construction of rock core images.

[0005] To achieve the above objectives, the present invention provides a method for color correction and profile construction of core images, comprising: Visible light sequence images of the same profile of the rock core under various controllable illuminations were obtained, and the intrinsic reflectance feature maps of the visible light sequence images were extracted after removing the geometric shadows and highlights of the illumination. By utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, reflectance-geometric consistency feature points of the rock core are extracted to construct a robust feature point set of the rock core. Construct the joint energy function of the visible light sequence image, combine the intrinsic reflectance feature map and the robust feature point set, and simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. Within the overlapping region of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map; The adaptive suture is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed, and the core digital profile image of the core is reconstructed.

[0006] Optionally, the intrinsic reflectance feature map of the visible light sequence image, stripped of illumination geometry, shadows, and highlights, is extracted, including: Subpixel-level registration is performed on each frame of the visible light sequence image to obtain a registered visible light sequence image. Calculate the initial normal vector and initial reflectance of each pixel on the surface of the registered visible light sequence image core; Using the consistency deviation between the initial reflectance and the grayscale values ​​of each frame of the registered visible light sequence image, the reprojection error of each pixel on the core surface under different lighting conditions is calculated, and the adaptive weight of each pixel on the core surface of the registered visible light sequence image is determined based on the reprojection error. Combining the adaptive weights, a weighted optimization objective function is constructed, and the initial normal vector and initial reflectivity are iteratively corrected until convergence, thereby outputting the intrinsic reflectivity feature map that has been stripped of the lighting geometry shadows and highlights.

[0007] Optionally, based on the reprojection error, the adaptive weights of each pixel on the surface of the registered visible light sequence image core are determined, including: Define robust control parameters for the registered visible light sequence images; Combining the reprojection error and the robust control parameters, the adaptive weights of each pixel on the core surface of the registered visible light sequence image are calculated using the following formula, including:

[0008] in, This indicates the registration of the visible light sequence image. The pixel at the th point Adaptive weights under individual light source observations This indicates the registration of the visible light sequence image. The actual observed gray level of each pixel in the visible light sequence image. This indicates the registration of the visible light sequence image. Initial reflectance of each pixel Indicates the first The unit incident light vector of a light source, This indicates the registration of the visible light sequence image. The initial normal vector of each pixel. Indicates robust control parameters, This indicates the registration of the visible light sequence image. The pixel at the th point Reprojection error under individual light source observation This represents an exponential function.

[0009] Optionally, by utilizing the texture distribution of the visible light sequence images and combining it with the intrinsic reflectance feature map, reflectance-geometric consistency feature points of the rock core are extracted, including: Based on the texture distribution and the intrinsic reflectance feature map, a joint confidence distribution map of the registered visible light sequence image corresponding to the visible light sequence image is generated; Based on the joint confidence distribution map, candidate feature points in the registered visible light sequence image that satisfy the preset texture intensity threshold and reflectivity stability threshold are marked; False feature points with unstable grayscale response under illumination changes are removed from the candidate feature points, and candidate feature points with high repetition rate and low illumination sensitivity are retained as the reflectivity-geometric consistency feature points.

[0010] Optionally, based on the texture distribution and the intrinsic reflectance feature map, a joint confidence distribution map of the registered visible light sequence image corresponding to the visible light sequence image is generated, including: Based on the texture distribution, the texture gradient response map of each frame in the registered visible light sequence image corresponding to the visible light sequence image is calculated to generate a texture saliency mask for each frame image; The color feature vectors of each pixel in the intrinsic reflectance feature map are extracted to construct a reflectance consistency constraint map that reflects the smoothness of the reflectance change on the rock core surface. The texture saliency mask and the reflectivity consistency constraint map are fused at the pixel level to generate the joint confidence distribution map.

[0011] Optionally, constructing a robust feature point set for the core includes: Sub-pixel level position refinement is performed on the reflectivity-geometric consistency feature points corresponding to the rock core to obtain sub-pixel feature points; For each of the sub-pixel feature points, a multi-scale description region with rotation invariance is constructed to generate a high-dimensional feature descriptor for the sub-pixel feature points; The sub-pixel feature points are matched using the high-dimensional feature descriptor to obtain matching results. Based on the matching results, the sub-pixel feature points are removed from the outlier, and the remaining correctly matched points are used as the final features to construct a robust feature point set for the core.

[0012] Optionally, the joint energy function of the visible light sequence images is constructed, including: Construct geometrical constraints for a robust feature point set corresponding to the visible light sequence image; Construct a color consistency constraint term corresponding to the intrinsic reflectance feature map of the visible light sequence image; Construct a transform field smoothing regularization term for the visible light sequence image; The geometric structure constraint, the color consistency constraint, and the transform field smoothing regularization term are weighted and summed to construct the joint energy function of the visible light sequence image.

[0013] Optionally, within the overlapping region of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map, including: Based on the pixel value distribution of the intrinsic reflectance feature map, the confidence value of each pixel in the overlapping area is calculated to generate a confidence distribution map corresponding to the spatial location of the intrinsic reflectance feature map. Within the overlapping area, using the confidence distribution map, a dynamic programming algorithm is employed to search for a path with the minimum total confidence cost as the initial stitching line, wherein the confidence cost is obtained by accumulating the confidence values ​​of each pixel on the path; The initial suture line is smoothed to obtain the final adaptive suture line.

[0014] Optionally, the adaptive suture line is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence decomposition to reconstruct the core digital profile image, including: Based on the adaptive suture line, a low-frequency fusion weight map of the low-frequency component and a high-frequency fusion weight map of the high-frequency component are generated respectively, so as to perform weighted fusion of the corresponding low-frequency components and high-frequency components of all frames in the intermediate image sequence to obtain fused low-frequency components and fused high-frequency components respectively. The low-frequency component and the high-frequency component are fused to reconstruct the digital profile image of the rock core.

[0015] To address the aforementioned problems, this invention also provides a core image color correction and profile construction system, the system comprising: The reflectance feature analysis module is used to acquire visible light sequence images of the same profile of the rock core under various controllable illuminations, and extract the intrinsic reflectance feature map of the visible light sequence images after removing the geometric shadows and highlights of the illumination. The robust feature point determination module is used to extract reflectance-geometric consistency feature points of the rock core by utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, so as to construct a robust feature point set of the rock core. The image correction and registration module is used to construct the joint energy function of the visible light sequence image, and combine the intrinsic reflectance feature map and the robust feature point set to simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. An adaptive suture analysis module is used to calculate an adaptive suture based on the intrinsic reflectance feature map within the overlapping area of ​​the intermediate image sequence. The digital profile image analysis module is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed using the adaptive suture line, and reconstruct the core digital profile image of the core. This invention acquires visible light sequence images of the same section of a rock core under different controllable illuminations and extracts its intrinsic reflectance feature map. This effectively eliminates the influence of illumination, geometric shadows, and highlights, fundamentally eliminating the image quality degradation caused by uneven illumination and surface undulations. This ensures that subsequent processing is based on stable and reliable material information. Using the texture distribution and intrinsic reflectance feature map of the visible light sequence images, a robust feature point set is constructed from the extracted reflectance-geometric consistency feature points. This significantly improves the stability and matching accuracy of the feature points, laying a solid foundation for subsequent accurate registration. By constructing a joint energy function, the color correction parameter matrix and the geometric registration transformation matrix are solved simultaneously. This invention achieves dual normalization of color and geometry, generating a highly consistent intermediate image sequence in both color and space. It solves the cumulative error problem caused by the separate processing of color and geometric registration in traditional methods. Within the overlapping areas of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map. This suture line is then used to perform reflectance confidence-weighted fusion of the decomposed high-frequency and low-frequency components, intelligently avoiding unreliable areas such as shadows and highlights, ensuring a smooth and natural fusion boundary. The resulting reconstructed digital profile image of the rock core is complete, seamless, and rich in detail, realistically reflecting the internal structure and mineral distribution of the core, providing high-quality and highly reliable visual evidence for geological analysis. Therefore, this invention can improve the accuracy of color correction and profile analysis of rock core images. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for color correction and profile construction of rock core images according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the modules for implementing the core image color correction and profile construction method according to an embodiment of the present invention; Figure 3 A schematic diagram of a computer device for a method of color correction and profile construction of rock core images provided in an embodiment of the present invention; The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0018] This application provides a method for color correction and profile construction of rock core images. The execution entity of the method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster.

[0019] Reference Figure 1 The diagram shown is a flowchart illustrating a core image color correction and profile construction method according to an embodiment of the present invention. In this embodiment, the core image color correction and profile construction method includes: S1. Obtain visible light sequence images of the same profile of the rock core under various controllable illuminations. Based on photometry and adaptive weight estimation, generate an intrinsic reflectance feature map stripped of the geometric shadows and highlights of the illumination.

[0020] It should be explained that the rock core refers to a cylindrical or semi-cylindrical rock sample obtained from underground through geological drilling; the same cross-section of the rock core refers to the cylindrical outer surface area of ​​the rock core extending along its axial direction; different controllable illuminations refer to lighting conditions with known differences in the illumination distribution pattern precisely controlled by the imaging system; and the visible light sequence image refers to a set of continuous images obtained by taking pictures of the same cross-section of the rock core under the different controllable illumination conditions using an image acquisition device. The image acquisition device can be an industrial area scan camera or a line scan camera.

[0021] This invention extracts the intrinsic reflectivity feature map of the visible light sequence image by stripping away the geometric shadow and highlight of the light, eliminating the light noise of the rock core that is not the object itself, and solving the color artifact problem caused by reflection and shadow in traditional single-light imaging. It provides a pure data foundation that is not affected by ambient light for subsequent accurate feature extraction and color correction of rock core based on physical properties.

[0022] Specifically, the extraction of the intrinsic reflectance feature map of the visible light sequence image, stripped of the illumination geometry, shadows, and highlights, includes: Subpixel-level registration is performed on each frame of the visible light sequence image to obtain a registered visible light sequence image; Calculate the initial normal vector and initial reflectance of each pixel on the surface of the registered visible light sequence image core; Using the consistency deviation between the initial reflectance and the grayscale values ​​of each frame of the registered visible light sequence image, the reprojection error of each pixel on the core surface under different lighting conditions is calculated, and the adaptive weight of each pixel on the core surface of the registered visible light sequence image is determined based on the reprojection error. Combining the adaptive weights, a weighted optimization objective function is constructed, and the initial normal vector and initial reflectivity are iteratively corrected until convergence, thereby outputting the intrinsic reflectivity feature map that has been stripped of the lighting geometry shadows and highlights.

[0023] The registered visible light sequence image refers to the image set obtained by spatially aligning each frame of the visible light sequence image in the pixel coordinate system using an image registration algorithm. The initial normal vector refers to the normal direction data of each pixel on the core surface, which is initially solved by the least squares method based on the basic assumption of photometric stereo vision and the linear relationship between the gray value of each pixel in the registered visible light sequence image and the known light source direction. The initial reflectivity refers to the inherent reflectivity coefficient of each pixel on the core surface to incident light, which is initially estimated by the illumination reflection model during the synchronous calculation of the initial normal vector. This coefficient reflects the mineral composition and color attributes of the rock itself. The consistency deviation refers to the degree of difference between the theoretically calculated grayscale value and the actual grayscale value acquired in the registered visible light sequence image. The reprojection error refers to the residual value obtained by comparing the theoretical brightness predicted after substituting the initial normal vector and initial reflectivity into the illumination reflection model with the actual observed brightness of the corresponding pixel in the registered visible light sequence image pixel by pixel. The brightness saturation region refers to the region in the registered visible light sequence image where the pixel grayscale value reaches or approaches the quantization limit of the imaging sensor, and the reprojection error is... The projection error exhibits a positive anomaly in certain regions. The low-brightness failure region refers to the region in the registered visible light sequence image where the pixel grayscale value is significantly lower than the theoretical expectation and the signal-to-noise ratio is extremely low. The adaptive weight refers to the numerical coefficients dynamically allocated based on the magnitude of the reprojection error calculated for each pixel under different lighting conditions. The weighted optimization objective function refers to the mathematical energy function used to describe the accuracy of the core surface normal vector and reflectance estimation. The intrinsic reflectance feature map refers to the grayscale image of the core surface reflectance output after minimizing the weighted optimization objective function and iteratively converging.

[0024] It should be noted that the aforementioned illumination reflection model refers to a mathematical equation describing the optical physical properties of the rock core surface. This model establishes a quantitative relationship among the observed gray value of a pixel in the visible light sequence image, the incident direction vector of the light source, the normal vector of the rock core surface, and the intrinsic reflectivity of the surface. The illumination reflection model is constructed based on the Lambert diffuse reflection theory. The pixel gray value being significantly lower than the theoretical expectation means that, under the current light source conditions, the standard brightness value that the rock core surface pixel should present when it is not shaded and the surface is fully reflective, calculated according to the illumination reflection model.

[0025] It should be explained that the weighted optimization objective function is based on confidence level error minimization, which aims to eliminate the interference of outlier data such as highlights and shadows on the calculation results through mathematical means. It transforms the standard least squares problem into a weighted least squares problem by introducing an adaptive weight matrix, thereby achieving differentiated control of the error contribution of different pixels.

[0026] Further, determining the adaptive weights of each pixel on the surface of the registered visible light sequence image core based on the reprojection error includes: Define robust control parameters for the registered visible light sequence images; Combining the reprojection error and the robust control parameters, the adaptive weights of each pixel on the core surface of the registered visible light sequence image are calculated using the following formula, including:

[0027] in, This indicates the registration of the visible light sequence image. The pixel at the th point Adaptive weights under individual light source observations This indicates the registration of the visible light sequence image. The actual observed gray level of each pixel in the visible light sequence image. This indicates the registration of the visible light sequence image. Initial reflectance of each pixel Indicates the first The unit incident light vector of a light source, This indicates the registration of the visible light sequence image. The initial normal vector of each pixel. Indicates robust control parameters, This indicates the registration of the visible light sequence image. The pixel at the th point Reprojection error under individual light source observation This represents an exponential function.

[0028] The robust control parameter refers to the parameter used to adjust the sensitivity of the adaptive weighting function to reprojection errors. The robust control parameter needs to be adaptively adjusted based on the texture complexity of the core surface, illumination intensity, and sensor noise level. The actual observed grayscale refers to the grayscale value of the core surface in the registered visible light sequence image. The pixel at the th point The brightness value actually captured and quantized by the image sensor under observation of a single light source. An exponential function is defined as an exponential function with the natural constant e as its base.

[0029] S2. Using the texture distribution of the visible light sequence image and the intrinsic reflectance feature map, extract the reflectance-geometric consistency feature points of the rock core to construct a robust feature point set of the rock core.

[0030] This invention utilizes the texture distribution of the visible light sequence image and combines it with the intrinsic reflectance feature map to extract the reflectance-geometric consistency feature point set of the rock core, which has both reflectance consistency and geometric structure stability. This significantly improves the accuracy and anti-interference ability of visible light sequence image feature matching in low-texture or repetitive texture areas.

[0031] In detail, the step of extracting reflectance-geometric consistency feature points from the rock core using the texture distribution of the visible light sequence image and the intrinsic reflectance feature map includes: Based on the texture distribution and the intrinsic reflectance feature map, a joint confidence distribution map of the registered visible light sequence image corresponding to the visible light sequence image is generated; Based on the joint confidence distribution map, candidate feature points in the registered visible light sequence image that satisfy the preset texture intensity threshold and reflectivity stability threshold are marked; False feature points with unstable grayscale response under illumination changes are removed from the candidate feature points, and candidate feature points with high repetition rate and low illumination sensitivity are retained as the reflectivity-geometric consistency feature points.

[0032] The texture distribution refers to the statistical regularity of the frequency and amplitude of pixel grayscale values ​​in the visible light sequence image in space. The joint confidence distribution map is a probability map generated by mathematically fusing the texture clarity information and reflectivity uniformity information of the rock core surface. The preset texture intensity threshold is the minimum texture gradient amplitude limit used to determine whether the pixels on the rock core surface have sufficient geometric details for feature matching. The reflectivity stability threshold is the maximum color change limit used to determine whether the reflectivity of the pixels on the rock core surface is uniform and unaffected by local highlights or shadows. The grayscale response is unstable under illumination changes. This refers to the physical phenomenon in the visible light sequence image where, with changes in the direction or intensity of illumination, the grayscale value of the same physical pixel fluctuates drastically, jumps abnormally, or changes nonlinearly between different frames. The pseudo-feature point refers to image noise or interference points among candidate feature points that, although exhibiting a high texture gradient response, lack illumination invariance. Specifically, these include highlight spots caused by the wetness of the rock surface, deep shadow edges caused by surface unevenness, and random noise points collected by the image sensor. The reflectivity-geometric consistency feature point refers to an image feature point that simultaneously satisfies the dual constraints of geometric texture saliency and intrinsic reflectivity stability.

[0033] Further, generating a joint confidence distribution map of the registered visible light sequence image corresponding to the visible light sequence image based on the texture distribution and the intrinsic reflectance feature map includes: Based on the texture distribution, the texture gradient response map of each frame in the registered visible light sequence image corresponding to the visible light sequence image is calculated to generate a texture saliency mask for each frame image; The color feature vectors of each pixel in the intrinsic reflectance feature map are extracted to construct a reflectance consistency constraint map that reflects the smoothness of the reflectance change on the rock core surface. The texture saliency mask and the reflectivity consistency constraint map are fused at the pixel level to generate the joint confidence distribution map.

[0034] Wherein, the texture gradient response map refers to the response matrix obtained after performing convolution operation on the registered visible light sequence image; the texture saliency mask refers to the binary image region generated by normalizing the values ​​of the texture gradient response map and performing binarization segmentation according to the preset texture intensity threshold; and the color feature vector refers to the multidimensional coordinate representation of each pixel in the intrinsic reflectance feature map in a specific color space.

[0035] This invention constructs a robust feature point set of the rock core and further filters reflectance-geometric consistency feature points, thereby improving the reliability of subsequent registration and correction of visible light sequence images.

[0036] Specifically, constructing the robust feature point set of the rock core includes: Sub-pixel level position refinement is performed on the reflectivity-geometric consistency feature points corresponding to the rock core to obtain sub-pixel feature points; For each of the sub-pixel feature points, a multi-scale description region with rotation invariance is constructed to generate a high-dimensional feature descriptor for the sub-pixel feature points; The sub-pixel feature points are matched using the high-dimensional feature descriptor to obtain matching results. Based on the matching results, the sub-pixel feature points are removed from the outlier, and the remaining correctly matched points are used as the final features to construct a robust feature point set for the core.

[0037] Wherein, the sub-pixel feature points refer to the set of high-precision coordinate points after improving the positioning accuracy of reflectance-geometric consistency feature points from the integer level to the sub-pixel level; the multi-scale description region refers to the image region centered on the sub-pixel feature points and adaptively selected according to the scale parameters of the feature points in the image Gaussian pyramid, which can contain the local main neighborhood structure of the feature points; the high-dimensional feature descriptor refers to the multi-dimensional numerical vector generated within the multi-scale description region by statistically analyzing the direction and magnitude distribution of the pixel gradients of the sub-pixel feature points, which can reflect the local texture structure, edge direction, and illumination robustness of the feature points; the matching result refers to the list of feature point pairs containing potential correspondences generated after calculating the similarity of the high-dimensional feature descriptors in different images; the correct matching point refers to the feature point pair among the sub-pixel feature points that conforms to the overall geometric transformation relationship between images and has a very high descriptor similarity; and the robust feature point set refers to the final feature point set formed by the convergence of the correct matching points.

[0038] Optionally, the sub-pixel level position refinement of the reflectivity-geometric consistency feature points corresponding to the rock core is performed to obtain the sub-pixel feature points by performing grayscale surface fitting or gradient interpolation operations on the reflectivity-geometric consistency feature points in the integer pixel neighborhood.

[0039] S3. Construct the joint energy function of the visible light sequence image, and combine the intrinsic reflectance feature map and the robust feature point set to simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is both color and geometrically normalized.

[0040] The present invention constructs a joint energy function for the visible light sequence image, enabling the simultaneous solution of the color correction parameter matrix and the geometric registration transformation matrix.

[0041] Specifically, the joint energy function for constructing the visible light sequence image includes: Construct geometrical constraints for a robust feature point set corresponding to the visible light sequence image; Construct a color consistency constraint term corresponding to the intrinsic reflectance feature map of the visible light sequence image; Construct a transform field smoothing regularization term for the visible light sequence image; The geometric structure constraint, the color consistency constraint, and the transform field smoothing regularization term are weighted and summed to construct the joint energy function of the visible light sequence image.

[0042] The geometric structure constraint term is used to measure the degree of overlap between the spatial position of the visible light sequence image after geometric transformation and the reference image, so as to constrain the geometric accuracy of image registration. The color consistency constraint term is used to constrain the solution of the color correction parameter matrix to ensure that color normalization is performed only on pixel regions with consistent materials. The transform field smoothing regularization term is used to suppress non-physical abrupt changes in the transform parameters on the image plane to ensure the spatial smoothness and continuity of the generated intermediate image sequence in terms of geometric deformation and color transition. The joint energy function refers to the global objective function obtained by linearly weighting the geometric structure constraint term, the color consistency constraint term and the transform field smoothing regularization term according to preset weight coefficients.

[0043] Optionally, the geometric constraint term for constructing the robust feature point set corresponding to the visible light sequence image is a mathematical function term constructed based on the coordinates of matching point pairs in the robust feature point set and using the principle of minimizing reprojection error.

[0044] Optionally, the color consistency constraint term corresponding to the intrinsic reflectance feature map of the visible light sequence image is constructed based on the pixel value distribution of the intrinsic reflectance feature map, and is an error term constructed using the reflectance difference between the visible light sequence image and the reference image at the same physical location.

[0045] Optionally, the transformation field smoothing regularization term constructed for the visible light sequence image refers to a penalty function term constructed based on the gradient of the geometric registration transformation matrix and the color correction parameter matrix in the spatial neighborhood.

[0046] It should be noted that the joint energy function includes:

[0047] in, Represents the joint energy function. Represents the geometric registration transformation matrix. Represents the color correction parameter matrix. Represents geometric constraint terms. The weights of the geometric constraint terms are represented. This indicates a color consistency constraint. This represents the weight of the color consistency constraint. This represents the smoothing regularization term of the transformed field. This represents the weight of the smoothing regularization term for the transformed field.

[0048] This invention combines the intrinsic reflectance feature map and the robust feature point set to simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image. This generates a joint optimization mechanism for intermediate image sequences with dual color and geometric normalization. This mechanism breaks the limitations of traditional serial processing modes that require correction before stitching or stitching before correction, ensuring dual convergence of overlapping areas in both color and geometric spaces. It avoids the phenomenon of color deviation misleading geometric alignment or geometric misalignment exacerbating color differences. The color correction parameter matrix is ​​used to perform nonlinear transformation on the pixel color space of the visible light sequence image. The set of parameters is designed to eliminate color inconsistencies in the visible light sequence images by compensating for differences in illumination, sensor spectral response, and ambient light color temperature between different image frames. The geometric registration transformation matrix is ​​used to describe the affine transformation of the spatial coordinate mapping relationship between each frame of the visible light sequence image and the reference image. The intermediate image sequence refers to a set of high-quality sequence images generated after the visible light sequence images are spatially geometrically aligned by the geometric registration transformation matrix and color normalized by the color correction parameter matrix, which eliminates both geometric position deviations and color differences.

[0049] S4. Within the overlapping area of ​​the intermediate image sequence, calculate the adaptive suture line based on the intrinsic reflectance feature map.

[0050] In the overlapping area of ​​the intermediate image sequence, the present invention calculates an adaptive stitching line based on the intrinsic reflectance feature map, which can intelligently avoid unreliable pixels caused by changes in illumination or missing information, ensuring that the stitching line falls in an area with consistent material and reliable information, and ultimately achieving visually seamless and geometrically continuous high-quality image fusion.

[0051] Specifically, calculating the adaptive suture line based on the intrinsic reflectance feature map within the overlapping region of the intermediate image sequence includes: Based on the pixel value distribution of the intrinsic reflectance feature map, the confidence value of each pixel in the overlapping area is calculated to generate a confidence distribution map corresponding to the spatial location of the intrinsic reflectance feature map. Within the overlapping area, using the confidence distribution map, a dynamic programming algorithm is employed to search for a path with the minimum total confidence cost as the initial stitching line, wherein the confidence cost is obtained by accumulating the confidence values ​​of each pixel on the path; The initial suture line is smoothed to obtain the final adaptive suture line.

[0052] Wherein, the overlapping region refers to the portion of the intermediate image sequence where any two adjacent frames share a common visible area in space; the confidence value refers to the reliability of the material information contained in a single pixel in the intrinsic reflectance feature map; the confidence distribution map refers to an image with the same spatial size as the intrinsic reflectance feature map; the dynamic programming algorithm refers to an algorithm used to find a path from one side of the overlapping region to the other side on the confidence distribution map that minimizes the total confidence cost; the initial stitching line refers to a path obtained by searching on the confidence distribution map using the dynamic programming algorithm; and the adaptive stitching line refers to a final path obtained after smoothing the initial stitching line.

[0053] It should be explained that the dynamic programming algorithm is implemented using Dijkstra's algorithm.

[0054] S5. Using the adaptive suture line, the high-frequency and low-frequency components after the intermediate image sequence decomposition are weighted and fused with reflectance confidence to reconstruct the core digital profile image.

[0055] This invention utilizes the adaptive suture line to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed. The reconstructed digital profile image of the rock core not only eliminates the seam traces in the overlapping area, but also effectively preserves the fine texture structure of the rock, ultimately reconstructing a high-fidelity, seamless, and geologically accurate digital profile image of the rock core.

[0056] Specifically, the step of using the adaptive suture line to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after decomposition of the intermediate image sequence to reconstruct the core digital profile image includes: Based on the adaptive suture line, a low-frequency fusion weight map of the low-frequency component and a high-frequency fusion weight map of the high-frequency component are generated respectively, so as to perform weighted fusion of the corresponding low-frequency components and high-frequency components of all frames in the intermediate image sequence to obtain the fused low-frequency component and the fused high-frequency component respectively. The low-frequency component and the high-frequency component are fused to reconstruct the digital profile image of the rock core.

[0057] The low-frequency fusion weight map indicates the contribution weight of the low-frequency components of each frame in the intermediate image sequence at the corresponding pixel during image fusion. The high-frequency fusion weight map indicates the contribution weight of the high-frequency components of each frame in the intermediate image sequence at the corresponding pixel during image fusion. The fused low-frequency component refers to a new low-frequency component image obtained by weighting and averaging the corresponding low-frequency components of all frames in the intermediate image sequence using the low-frequency fusion weight map. The fused high-frequency component refers to a new high-frequency component image obtained by weighting and averaging the corresponding high-frequency components of all frames in the intermediate image sequence using the high-frequency fusion weight map. The core digital profile image refers to the final panoramic image obtained by synthesizing the fused low-frequency component and the fused high-frequency component.

[0058] This invention acquires visible light sequence images of the same section of a rock core under different controllable illuminations and extracts its intrinsic reflectance feature map. This effectively eliminates the influence of illumination, geometric shadows, and highlights, fundamentally eliminating the image quality degradation caused by uneven illumination and surface undulations. This ensures that subsequent processing is based on stable and reliable material information. Using the texture distribution and intrinsic reflectance feature map of the visible light sequence images, a robust feature point set is constructed from the extracted reflectance-geometric consistency feature points. This significantly improves the stability and matching accuracy of the feature points, laying a solid foundation for subsequent accurate registration. By constructing a joint energy function, the color correction parameter matrix and the geometric registration transformation matrix are solved simultaneously. This invention achieves dual normalization of color and geometry, generating a highly consistent intermediate image sequence in both color and space. It solves the cumulative error problem caused by the separate processing of color and geometric registration in traditional methods. Within the overlapping areas of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map. This suture line is then used to perform reflectance confidence-weighted fusion of the decomposed high-frequency and low-frequency components, intelligently avoiding unreliable areas such as shadows and highlights, ensuring a smooth and natural fusion boundary. The resulting reconstructed digital profile image of the rock core is complete, seamless, and rich in detail, realistically reflecting the internal structure and mineral distribution of the core, providing high-quality and highly reliable visual evidence for geological analysis. Therefore, this invention can improve the accuracy of color correction and profile analysis of rock core images.

[0059] like Figure 2 The diagram shown is a functional module diagram of the rock core image color correction and profile construction system of the present invention.

[0060] The core image color correction and profile construction system 200 of this invention can be installed in an electronic device. Depending on the functions implemented, the core image color correction and profile construction system may include a reflectance feature analysis module 201, a robust feature point determination module 202, an image correction and registration module 203, an adaptive suture analysis module 204, and a digital profile image analysis module 205. The module described in this invention can also be referred to as a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0061] In this embodiment of the invention, the functions of each module / unit are as follows: The reflectance feature analysis module 201 is used to acquire visible light sequence images of the same profile of the rock core under various different controllable illuminations, and extract the intrinsic reflectance feature map of the visible light sequence images after stripping the geometric shadows and highlights of the illumination. The robust feature point determination module 202 is used to extract reflectance-geometric consistency feature points of the rock core by utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, so as to construct a robust feature point set of the rock core. The image correction and registration module 203 is used to construct the joint energy function of the visible light sequence image, and combine the intrinsic reflectance feature map and the robust feature point set to simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. The adaptive suture analysis module 204 is used to calculate an adaptive suture based on the intrinsic reflectance feature map within the overlapping area of ​​the intermediate image sequence. The digital profile image analysis module 205 is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed using the adaptive suture line to reconstruct the core digital profile image.

[0062] In detail, the modules in the core image color correction and profile construction system 200 described in this embodiment of the invention employ the same methods as described above. Figure 1 The core image color correction and profile construction methods described herein use the same techniques and can produce the same technical effects, so they will not be elaborated here.

[0063] In one embodiment, a computer device is provided, which may be a server or a client, and its internal structure diagram may be as follows: Figure 3As shown, the computer device includes a processor, memory, network interface, and database connected via a system bus. The processor provides computational and control capabilities. The memory includes non-volatile and / or volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The network interface is used for communication with external clients via a network connection. When the computer program is executed by the processor, it implements the functions or steps of a core image color correction and profile construction method on the server or client side.

[0064] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to perform the following steps: Visible light sequence images of the same profile of the rock core under various controllable illuminations were obtained, and the intrinsic reflectance feature maps of the visible light sequence images were extracted after removing the geometric shadows and highlights of the illumination. By utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, reflectance-geometric consistency feature points of the rock core are extracted to construct a robust feature point set of the rock core. Construct the joint energy function of the visible light sequence image, combine the intrinsic reflectance feature map and the robust feature point set, and simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. Within the overlapping region of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map; The adaptive suture is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed, and the core digital profile image of the core is reconstructed.

[0065] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, the computer program performing the following steps when executed by a processor: Visible light sequence images of the same profile of the rock core under various controllable illuminations were obtained, and the intrinsic reflectance feature maps of the visible light sequence images were extracted after removing the geometric shadows and highlights of the illumination. By utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, reflectance-geometric consistency feature points of the rock core are extracted to construct a robust feature point set of the rock core. Construct the joint energy function of the visible light sequence image, combine the intrinsic reflectance feature map and the robust feature point set, and simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. Within the overlapping region of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map; The adaptive suture is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed, and the core digital profile image of the core is reconstructed.

[0066] It should be noted that the functions or steps that can be implemented by the computer-readable storage medium or computer device described above can be referred to the relevant descriptions on the server side and client side in the foregoing method embodiments. To avoid repetition, they will not be described one by one here.

[0067] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0068] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0069] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0070] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for color correction and profile construction of rock core images, characterized in that, The method includes: Visible light sequence images of the same profile of the rock core under various controllable illuminations were obtained, and the intrinsic reflectance feature maps of the visible light sequence images were extracted after removing the geometric shadows and highlights of the illumination. By utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, reflectance-geometric consistency feature points of the rock core are extracted to construct a robust feature point set of the rock core. Construct the joint energy function of the visible light sequence image, combine the intrinsic reflectance feature map and the robust feature point set, and simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. Within the overlapping region of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map; The adaptive suture is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed, and the core digital profile image of the core is reconstructed.

2. The method for color correction and profile construction of core images as described in claim 1, characterized in that, The intrinsic reflectance feature maps of the visible light sequence images, stripped of illumination geometry, shadows, and highlights, are extracted, including: Subpixel-level registration is performed on each frame of the visible light sequence image to obtain a registered visible light sequence image. Calculate the initial normal vector and initial reflectance of each pixel on the surface of the registered visible light sequence image core; Using the consistency deviation between the initial reflectance and the grayscale values ​​of each frame of the registered visible light sequence image, the reprojection error of each pixel on the core surface under different lighting conditions is calculated, and the adaptive weight of each pixel on the core surface of the registered visible light sequence image is determined based on the reprojection error. Combining the adaptive weights, a weighted optimization objective function is constructed, and the initial normal vector and initial reflectivity are iteratively corrected until convergence, thereby outputting the intrinsic reflectivity feature map that has been stripped of the lighting geometry shadows and highlights.

3. The method for color correction and profile construction of core images as described in claim 2, characterized in that, Based on the reprojection error, the adaptive weights of each pixel on the surface of the registered visible light sequence image core are determined, including: Define robust control parameters for the registered visible light sequence images; Combining the reprojection error and the robust control parameters, the adaptive weights of each pixel on the core surface of the registered visible light sequence image are calculated using the following formula, including: ; in, This indicates the registration of the visible light sequence image. The pixel at the th point Adaptive weights under individual light source observations This indicates the registration of the visible light sequence image. The actual observed gray level of each pixel in the visible light sequence image. This indicates the registration of the visible light sequence image. Initial reflectance of each pixel Indicates the first The unit incident light vector of a light source, This indicates the registration of the visible light sequence image. The initial normal vector of each pixel. Indicates robust control parameters, This indicates the registration of the visible light sequence image. The pixel at the th point Reprojection error under individual light source observation This represents an exponential function.

4. The method for color correction and profile construction of core images as described in claim 1, characterized in that, Using the texture distribution of the visible light sequence images and combining it with the intrinsic reflectance feature map, the reflectance-geometric consistency feature points of the rock core are extracted, including: Based on the texture distribution and the intrinsic reflectance feature map, a joint confidence distribution map of the registered visible light sequence image corresponding to the visible light sequence image is generated; Based on the joint confidence distribution map, candidate feature points in the registered visible light sequence image that satisfy the preset texture intensity threshold and reflectivity stability threshold are marked; False feature points with unstable grayscale response under illumination changes are removed from the candidate feature points, and candidate feature points with high repetition rate and low illumination sensitivity are retained as the reflectivity-geometric consistency feature points.

5. The method for color correction and profile construction of core images as described in claim 4, characterized in that, Based on the texture distribution and the intrinsic reflectance feature map, a joint confidence distribution map of the registered visible light sequence image corresponding to the visible light sequence image is generated, including: Based on the texture distribution, the texture gradient response map of each frame in the registered visible light sequence image corresponding to the visible light sequence image is calculated to generate a texture saliency mask for each frame image; The color feature vectors of each pixel in the intrinsic reflectance feature map are extracted to construct a reflectance consistency constraint map that reflects the smoothness of the reflectance change on the rock core surface. The texture saliency mask and the reflectivity consistency constraint map are fused at the pixel level to generate the joint confidence distribution map.

6. The method for color correction and profile construction of core images as described in claim 1, characterized in that, Constructing a robust feature point set for the core includes: Sub-pixel level position refinement is performed on the reflectivity-geometric consistency feature points corresponding to the rock core to obtain sub-pixel feature points; For each of the sub-pixel feature points, a multi-scale description region with rotation invariance is constructed to generate a high-dimensional feature descriptor for the sub-pixel feature points; The sub-pixel feature points are matched using the high-dimensional feature descriptor to obtain matching results. Based on the matching results, the sub-pixel feature points are removed from the outlier, and the remaining correctly matched points are used as the final features to construct a robust feature point set for the core.

7. The method for color correction and profile construction of core images as described in claim 1, characterized in that, Constructing the joint energy function of the visible light sequence images includes: Construct geometrical constraints for a robust feature point set corresponding to the visible light sequence image; Construct a color consistency constraint term corresponding to the intrinsic reflectance feature map of the visible light sequence image; Construct a transform field smoothing regularization term for the visible light sequence image; The geometric structure constraint, the color consistency constraint, and the transform field smoothing regularization term are weighted and summed to construct the joint energy function of the visible light sequence image.

8. The method for color correction and profile construction of core images as described in claim 1, characterized in that, Within the overlapping region of the intermediate image sequence, an adaptive suture line is calculated based on the intrinsic reflectance feature map, including: Based on the pixel value distribution of the intrinsic reflectance feature map, the confidence value of each pixel in the overlapping area is calculated to generate a confidence distribution map corresponding to the spatial location of the intrinsic reflectance feature map. Within the overlapping area, using the confidence distribution map, a dynamic programming algorithm is employed to search for a path with the minimum total confidence cost as the initial stitching line, wherein the confidence cost is obtained by accumulating the confidence values ​​of each pixel on the path; The initial suture line is smoothed to obtain the final adaptive suture line.

9. The method for color correction and profile construction of core images as described in claim 8, characterized in that, The adaptive suture line is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after decomposition of the intermediate image sequence to reconstruct the digital profile image of the core, including: Based on the adaptive suture line, a low-frequency fusion weight map of the low-frequency component and a high-frequency fusion weight map of the high-frequency component are generated respectively, so as to perform weighted fusion of the corresponding low-frequency components and high-frequency components of all frames in the intermediate image sequence to obtain the fused low-frequency component and the fused high-frequency component respectively. The low-frequency component and the high-frequency component are fused to reconstruct the digital profile image of the rock core.

10. A core image color correction and profile construction system, characterized in that, The system includes: The reflectance feature analysis module is used to acquire visible light sequence images of the same profile of the rock core under various controllable illuminations, and extract the intrinsic reflectance feature map of the visible light sequence images after removing the geometric shadows and highlights of the illumination. The robust feature point determination module is used to extract reflectance-geometric consistency feature points of the rock core by utilizing the texture distribution of the visible light sequence image and combining it with the intrinsic reflectance feature map, so as to construct a robust feature point set of the rock core. The image correction and registration module is used to construct the joint energy function of the visible light sequence image, and combine the intrinsic reflectance feature map and the robust feature point set to simultaneously solve the color correction parameter matrix and geometric registration transformation matrix of the visible light sequence image to generate an intermediate image sequence that is dually normalized in color and geometry. An adaptive suture analysis module is used to calculate an adaptive suture based on the intrinsic reflectance feature map within the overlapping area of ​​the intermediate image sequence. The digital profile image analysis module is used to perform reflectance confidence-weighted fusion of the high-frequency and low-frequency components after the intermediate image sequence is decomposed using the adaptive suture line, and reconstruct the core digital profile image of the core.

Images