Method and device for quickly generating sketches based on real-life photos of rock cores

Abstract

The present invention relates to a method and device for quickly generating a sketch based on a real-life photo of a core, comprising: obtaining a photo of a core box, and rotating the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm, so as to detect the range of the corresponding core grid in the photo of the core box, and obtain the core range. The core grid is cropped based on the input cropping boundary parameter, and the core cropping parameter is obtained, and the corresponding core box bottom depth value in the core cropping parameter is attached to each cropped core grid. The core grids whose core recovery rate exceeds a first threshold are directly spliced ​​to form a sketch according to the well depth or footage, and the local core missing section of the core grid whose core recovery rate is lower than the first threshold is stretched and spliced ​​into a sketch. A 3D scene library is used to construct a preview scene for the storage of a core sketch bar chart, and the sketch formed by the splicing of the core grids and the cataloging results are visually displayed, with a high degree of intelligence.

Description

Technical Field

[0001] The present invention relates to the technical field of geotechnical exploration, and in particular to a method and device for rapidly generating a sketch map based on a real-life photograph of a rock core. Background Art

[0002] In the field of geotechnical exploration, drilling core sampling in order to explore the lithology of the proposed engineering site is an important part of ascertaining the geological conditions. The cylindrical core samples obtained from the geological exploration boreholes can intuitively display the development of structural surface fissures, stratum weathering boundaries, geological sedimentation laws, stratum structure and other geological information. Usually, the drilled cores will be neatly placed in the grid of the core box according to the depth, and then the drillers will take photos after placing the core box neatly according to the depth on site. After the exploration task is completed, the drilling core cataloging results and core photos are the necessary basis for drilling acceptance and review, and are also important content for the reviewers to evaluate the quality of on-site exploration.

[0003] At present, the industry still uses the traditional operation method to carry out geotechnical survey work, that is, professional technicians conduct simple classification of core samples on site, and then record the geological information reflected by the drilling core sampling and the in-situ test data and other cataloging results in the form of paper operations, and finally arrange the core boxes neatly and take photos for archiving. However, this manual and paper-based operation method consumes a lot of human resources and time. At the same time, the messy working environment on site and the uneven professional level of technicians will greatly affect the results of drilling cataloging, and will also affect the evaluation results of the exploration quality by the auditors in the next link. When faced with a large number of exploration tasks, the work intensity of field workers is high, and the work time required for hand-drawn sketches is long, which can easily cause work errors, low work efficiency, and difficult to ensure work quality, affecting the overall progress of the project. In addition, the field supervision of the exploration site is difficult, and the uploaded paper exploration results are seriously falsified. A large amount of paper materials and various manual hand-drawn sketches make it difficult to ensure accurate control of the exploration quality. Summary of the invention

[0004] Based on this, it is necessary to provide a method and device for quickly generating sketches based on core real-life photos, which can intelligently generate core real-life sketches with high efficiency and high exploration quality accuracy, in order to address the above technical problems.

[0005] The present invention provides a method for quickly generating a sketch based on a core real-life photograph, the method comprising:

[0006] Obtain a core box photo, and rotate the core box photo to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range;

[0007] The core grid is trimmed based on the input trimming boundary parameters, and the core trimming parameters are obtained, and the core box bottom depth value corresponding to the core trimming parameters is attached to each of the trimmed core grids;

[0008] According to the well depth or footage, the core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​into the sketch map to obtain a core sketch column map;

[0009] A preview scene is constructed for the storage of the core sketch histogram using a 3D scene library, and the preview scene is used to visualize the sketch formed by splicing the core grid and the cataloging results;

[0010] The core cutting parameters include the core box length value, the number of core grids and the core box bottom depth value.

[0011] Further, the core box photo is obtained, and the core box photo is rotated to a front view angle by an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range, including:

[0012] Capturing the core box photo, and setting a Matrix operation matrix of the picture after capturing the core box photo, to obtain a core box picture matrix;

[0013] The core box photo is scaled by setting the x and y directions in the core box photo matrix and the scale ratio is set.

[0014] Further, the core box photo is obtained, and the core box photo is rotated to a front view angle by an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range, and further includes:

[0015] Based on the scaled core box photo, detecting the rotation angle of the core box photo, and calling ExifInterface to obtain the rotation angle of the core box photo according to the information carried by the core box photo;

[0016] According to the rotation angle of the core box photograph, the core box photograph is rotated to a normal viewing angle with the axis of the x-axis and the y-axis as a reference point.

[0017] Further, the core box photo is obtained, and the core box photo is rotated to a front view angle by an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range, and further includes:

[0018] The core range in the core box photo is detected by using a single-class object instance segmentation method in a Mask-RCNN instance segmentation algorithm, and a Mask R-CNN is set for pixel points of the core object within the core range;

[0019] The Mask-RCNN instance segmentation model is trained using a plurality of different core box photos for determining the core range to obtain a pre-trained neural network.

[0020] Further, the core box photo is obtained, and the core box photo is rotated to a front view angle by an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range, and further includes:

[0021] Using the core box photo to be detected as the input of the pre-trained neural network, so as to call the pre-trained neural network to perform convolution and pooling processing on the pixel points of the core box photo, and obtain the prediction result of the core range;

[0022] The prediction result is compared with the core box photo to be detected to obtain an alignment deviation value between the pixels of the prediction result and the core box photo to be detected, and a mask for setting the core pixel points is calculated by bilinear interpolation so that the alignment deviation value reaches a second threshold.

[0023] Furthermore, the core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map according to the well depth or footage, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​to form the sketch map, so as to obtain a core sketch column map, including:

[0024] Detecting whether there is a local core missing section at the break of the taken core, and if there is a local core missing section at the break of the taken core, longitudinally stretching the missing section of the single core according to a preset depth so that the single core reaches the actual depth position;

[0025] The starting position and the ending position of the missing section of the single core are marked to determine the core interval to be stretched, and the core interval is longitudinally stretched according to the actual depth so that each core reaches the actual depth position.

[0026] Furthermore, the method further comprises:

[0027] The 3D scene library is used to create a printing scene for rendering graphic content, wherein the printing scene uses configuration options in a renderer WebGLRenderer to generate a graphics buffer;

[0028] Through the printing scene, each page of the preview scene is rendered to the canvas according to the printing paper height and the drilling length, and the rendering content includes a table header, grid lines, structured data annotations, and a sketch formed by splicing the current page in the order of the core box bottom depth values.

[0029] The present invention also provides a device for quickly generating a sketch based on a real-life photograph of a core, the device comprising:

[0030] The core range detection module is used to obtain a core box photo and rotate the core box photo to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range;

[0031] A core grid cutting module is used to cut the core grid based on the input cutting boundary parameters, obtain the core cutting parameters, and attach the core box bottom depth value corresponding to the core cutting parameters to each of the cut core grids;

[0032] A core grid splicing module is used to directly splice the core grids whose core recovery rate exceeds the first threshold value according to the well depth or footage to form a sketch map, and to stretch the local core missing sections of the core grids whose core recovery rate is lower than the first threshold value and then splice them into the sketch map to obtain a core sketch column map;

[0033] A visualization display module, for constructing a preview scene for the storage of the core sketch histogram using a 3D scene library, wherein the preview scene is used for visually displaying the sketch formed by splicing the core grid and the cataloging results;

[0034] The core cutting parameters include the core box length value, the number of core grids and the core box bottom depth value.

[0035] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, it implements any of the above-mentioned methods for quickly generating a sketch based on a real-life photograph of a core.

[0036] The present invention also provides a computer storage medium storing a computer program, wherein when the computer program is executed by a processor, the method for quickly generating a sketch based on a real-life photograph of a core as described above is implemented.

[0037] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for quickly generating a sketch based on a real-life photograph of a core.

[0038] The above-mentioned method and device for quickly generating sketches based on real-life photos of cores, by introducing a mature artificial intelligence graphic recognition algorithm, can intelligently detect the angle and range of the required core cutting through the core photos taken at the construction site. After setting the cutting parameters of the core photos, the core photos are cut into single core grids in equal proportions, and automatically spliced ​​according to the depth to form a drilling core sketch. Then, the drilling geological information is collected through the structured data cataloging form in the system, and the cataloging results are directly marked on the drilling core sketch according to the depth, including a detailed description of the geological structure of the core, in-situ test data, sampling information, and structured data such as groundwater level. The electronic collection method also stipulates a unified standard for the placement of cores and taking photos, supports the field exploration operation mode of paperless office, abandons the errors and low work efficiency caused by manual hand-drawing, improves the efficiency of field work, and can also play a role in field supervision through big data traces, which can effectively prevent the falsification of drilling cataloging results. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0040] Figure 1 One of the schematic diagrams of the process of the method for quickly generating a sketch based on a real-life photo of a core provided by the present invention;

[0041] Figure 2 A second schematic flow diagram of a method for quickly generating a sketch based on a core real-life photo provided by the present invention;

[0042] Figure 3 A third schematic flow diagram of a method for quickly generating a sketch based on a core real-life photo provided by the present invention;

[0043] Figure 4 A fourth flow chart of a method for quickly generating a sketch based on a core real-life photo provided by the present invention;

[0044] Figure 5 A fifth flow chart of a method for quickly generating a sketch based on a core real-life photo provided by the present invention;

[0045] Figure 6 Schematic diagram of the sixth process of the method for quickly generating a sketch based on a core real-life photo provided by the present invention;

[0046] Figure 7 Schematic diagram of the seventh process of the method for quickly generating a sketch based on a core real-life photo provided by the present invention;

[0047] Figure 8 A schematic diagram of the overall process of sketch generation and visual rendering of a method for quickly generating sketches based on core real-life photos in a specific embodiment provided by the present invention;

[0048] Fig. 9 A schematic diagram of the structure of a device for quickly generating a sketch based on a real-life photo of a core provided by the present invention;

[0049] Fig.10 This is a diagram of the internal structure of the computer device provided by the present invention. DETAILED DESCRIPTION

[0050] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] Combine the following Figure 1-Figure 10 The invention describes a method and device for quickly generating a sketch map based on a real-life photograph of a rock core.

[0052] like Figure 1 As shown, in one embodiment, a method for quickly generating a sketch based on a core real-life photograph comprises the following steps:

[0053] Step S110, obtaining a core box photo, and rotating the core box photo to a front view angle through an artificial intelligence graphic recognition algorithm, so as to detect the range of the corresponding core grid in the core box photo and obtain the core range.

[0054] Step S120, cutting the core grid based on the input cutting boundary parameters, and obtaining the core cutting parameters, and attaching the core box bottom depth value corresponding to the core cutting parameters to each cut core grid.

[0055] Step S130, directly splicing the core grids whose core recovery rate exceeds the first threshold value into a sketch map according to the well depth or footage, and stretching the local core missing sections of the core grids whose core recovery rate is lower than the first threshold value and then splicing them into a sketch map to obtain a core sketch column map.

[0056] The first threshold is a set threshold.

[0057] Step S140, using the 3D scene library to construct a preview scene for the storage of the core sketch bar graph, the preview scene is used to visualize the sketch graph formed by splicing the core grid and the cataloging results.

[0058] The above method of quickly generating sketches based on real-life photos of cores can intelligently detect the angle and range of the required core cutting through the core photos taken at the construction site by introducing a mature artificial intelligence graphic recognition algorithm. After setting the cutting parameters of the core photos, the core photos are cut into single core grids in equal proportions, and automatically spliced ​​according to the depth to form a sketch of the drilled core. Then, the drilling geological information is collected through the structured data cataloging form in the system, and the cataloging results are directly marked on the sketch of the drilled core according to the depth, including a detailed description of the geological structure of the core, in-situ test data, sampling information, and structured data such as groundwater level. The electronic collection method also stipulates a unified standard for the placement of cores and taking photos, supports the field exploration operation mode of paperless office, abandons the errors and low work efficiency caused by manual hand-drawing, improves the efficiency of field work, and can also play a role in field supervision through big data traces, which can effectively prevent the falsification of drilling cataloging results.

[0059] like Figure 2 As shown, in one embodiment, the method for quickly generating a sketch based on a real-life photo of a core provided by the present invention obtains a photo of a core box, and rotates the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the photo of the core box, and obtain the core range, which specifically includes the following steps:

[0060] Step S111, capturing the core box photo, and setting the Matrix operation matrix of the photo after capturing the core box photo, to obtain the core box photo matrix.

[0061] Among them, the core box image matrix is ​​a Matrix type operation matrix.

[0062] Step S112, scaling the core box photo by setting the x and y directions in the core box photo matrix and using the set scaling ratio.

[0063] like Figure 3 As shown, in one embodiment, the method for quickly generating a sketch based on a real-life photo of a core provided by the present invention obtains a photo of a core box, and rotates the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the photo of the core box, and obtain the core range, which specifically includes the following steps:

[0064] Step S113, based on the scaled core box photo, detecting the rotation angle of the core box photo, and calling ExifInterface to obtain the rotation angle of the core box photo according to the information carried by the core box photo.

[0065] Among them, in Android multimedia development, ExifInterface (Exif exchangeable image file) is an interface that provides information such as rotation, GPS, and time of image files.

[0066] Step S114, according to the rotation angle of the core box photograph, the core box photograph is rotated to a normal viewing angle with the axis of the x-axis and the y-axis as a reference point.

[0067] like Figure 4 As shown, in one embodiment, the method for quickly generating a sketch based on a real-life photo of a core provided by the present invention obtains a photo of a core box, and rotates the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the photo of the core box, and obtain the core range, which specifically includes the following steps:

[0068] Step S115, detecting the core range in the core box photo by using the single-class object instance segmentation method in the Mask-RCNN instance segmentation algorithm, and setting a Mask R-CNN for the pixel points of the core object within the core range.

[0069] Step S116, training the Mask-RCNN instance segmentation model through a plurality of different core box photos for determining the core range, to obtain a pre-trained neural network.

[0070] like Figure 5 As shown, in one embodiment, the method for quickly generating a sketch based on a real-life photo of a core provided by the present invention obtains a photo of a core box, and rotates the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the photo of the core box, and obtain the core range, which specifically includes the following steps:

[0071] Step S117, using the core box photo to be detected as the input of the pre-trained neural network, so as to call the pre-trained neural network to perform convolution and pooling processing on the pixel points of the core box photo, and obtain the prediction result of the core range.

[0072] Step S118, comparing the prediction result with the core box photo to be detected, obtaining an alignment deviation value between the pixels of the prediction result and the core box photo to be detected, and calculating a mask for setting the core pixel points by bilinear interpolation so that the alignment deviation value reaches a second threshold.

[0073] The second threshold is a preset threshold or the number of iteration steps reaches a preset value.

[0074] like Figure 6 As shown, in one embodiment, the method for quickly generating a sketch map based on a core real-life photo provided by the present invention directly splices the core grids whose core recovery rate exceeds a first threshold value according to the well depth or footage to form a sketch map, and stretches the local core missing sections of the core grids whose core recovery rate is lower than the first threshold value and splices them into a sketch map to obtain a core sketch column map, which specifically includes the following steps:

[0075] Step S131, detecting whether there is a local core missing section at the break of the taken core, and if there is a local core missing section at the break of the taken core, longitudinally stretching the missing section of the single core according to a preset depth so that the single core reaches the actual depth position.

[0076] Step S132, marking the starting position and the ending position of the missing section of a single core to determine the core interval to be stretched, and longitudinally stretching the core interval according to the actual depth so that each core reaches the actual depth position.

[0077] like Figure 7 As shown, in one embodiment, the method for quickly generating a sketch based on a core real-life photo provided by the present invention further includes the following steps:

[0078] Step S710: Use a 3D scene library to create a printing scene for rendering graphic content. The printing scene uses configuration options in a renderer WebGLRenderer to generate a graphics buffer.

[0079] Among them, WebGLRenderer is the renderer category.

[0080] Step S720, render each page of the preview scene to the canvas according to the printing paper height and the drilling length through the printing scene, and the rendering content includes a table header, grid lines, structured data annotations, and a sketch formed by splicing the current page in the order of the core box bottom depth values.

[0081] See also Figure 8 As shown, in a specific embodiment, the method for quickly generating a sketch based on a core real-life photo provided by the present invention first requires inputting a core box photo, rotating the core box photo to a front view angle through an artificial intelligence graphic recognition algorithm, and detecting the range of the core grid. After the system captures the core box photo, it sets the Matrix operation matrix of the picture, and scales the core box photo by setting the x and y directions in the picture matrix and at a scaling ratio of 0.5 float (50%).

[0082] In this embodiment, the rotation angle of the image of the core box is detected, and the rotation angle of the image (0°, 90°, 180°, 270°) is obtained according to the information carried by the image file using ExifInterface; according to the obtained rotation angle, the XY axis center is used as the reference point for rotation until the image is corrected (0° will not rotate). The core range in the core box is detected by the (single-class object) instance segmentation method in the Mask-RCNN instance segmentation algorithm, and a mask R-CNN is set for the pixel points of the core object in the sample. Through deep learning and training of a large number of core samples, a pre-trained neural network is obtained, and then the core photo to be processed is input, and the pixel points of the graphic are convoluted and pooled, so that the size of the core photo is continuously reduced, and a prediction result of the core area range is obtained, which is compared with the input image to obtain the alignment deviation value between pixels; then a deconvolution operation is performed, that is, bilinear interpolation is used to calculate the characteristic value of the core (the mask set by the core pixel point), until the deviation value reaches the preset threshold or the number of iteration steps reaches the preset value, and the area range detection of each core grid in the core box is completed.

[0083] It should be noted that convolution: extracts local features of input data through convolution operation, that is, extracts core feature values ​​(masks set by core pixel points) through operation. Pooling: uses convolutional neural networks to detect target objects in graphic objects. The purpose is to perform maximum pooling on objects of non-uniform size to obtain a fixed-size feature map, which is used to detect the area range of multiple cores in the core box photo.

[0084] After determining the core range, input the cutting boundary parameters to cut the grid of the core box in equal proportion and store it in the system database. Set the core cutting parameters in the system, including the core box bottom depth, core box length, and grid number. On this basis, cut the detected core range in equal proportion and attach the depth value to each core grid after cutting.

[0085] Subsequently, the cut single core grids are processed. For single grids with high core recovery rates (generally referring to a recovery rate greater than 95%), they are directly spliced ​​according to the well depth or footage to form a sketch. For single grid cores with low core recovery rates (generally referring to less than 95%), it is necessary to first stretch the local core missing sections (for example, core missing due to mechanical wear and other human factors) and then splice them.

[0086] In this embodiment, when it is found that the middle part of the taken core is broken and there is a missing section, the missing section of the single core needs to be longitudinally stretched according to the preset depth to reach the actual depth position. First, the starting and ending positions of the missing section need to be marked to determine the core interval that needs to be processed, and then the interval is longitudinally stretched according to the actual depth to achieve the effect that the whole single core meets the actual length.

[0087] In this embodiment, the system uses three.js (3D software library) to build two scenes for the storage of core sketch histograms, a preview scene and a print scene. The preview scene is rendered as a user-visible area for the display of core splicing and cataloging results, which can be dragged and scaled. The print scene is rendered on the page and hidden in the system background. The print scene uses the configuration option buffer (preserveDrawingBuffer) of the renderer (WebGLRenderer) to generate a graphics buffer, render the content of the preview scene to canvas (canvas, an area for displaying rendering results) and convert it to generate a PDF file of A4 paper size. Since this option configuration consumes more memory and for performance considerations, the rendering does not need to be real-time rendering, but is instead rendered on demand to reduce the number of cached frames.

[0088] Among them, 2 scenarios include:

[0089] (1) The system creates a plane rectangular object in the preview scene (plane is a rectangular model object in the 3D scene. By setting the U and V coordinates, the stretched core grid image object is pasted onto the rectangle, and then the core grids are stitched in order according to the depth information attached to each core grid. The stitched result is placed at the specified position of the preview scene.) The proportionally cut core grid strips are pasted onto the texture UV coordinates. The lower left corner of the image is the coordinate origin (0,0), and the upper right corner is the coordinate (1.0,1.0). The vertical and horizontal coordinates of all positions on the image are between 0.0 and 1.0. The stretched single core strip is pasted onto the rectangle in proportion to ensure that each cut core grid is completely mapped onto the plane rectangular object. Then, the grids can be stitched together to form a sketch bar chart of the drill core displayed by depth.

[0090] (2) The system prints the scene using Line (Line is a line object in the 3D scene, used to draw the table style and grid (1cm×1cm grid paper) in the print scene) to create table and grid objects, draw the table according to the A4 paper aspect ratio and table content style, and use css2d (used to create text and place them as 2D objects at specific locations in the 3D scene. These elements will maintain the correct screen space projection as the camera moves and will not be affected by 3D transformations. They are used to create annotations or information prompts for the print scene, etc.) to add text annotations. The text is marked at a fixed position according to the depth based on the drilling catalog results and is scaled proportionally.

[0091] For example, the system uses the 3D scene library three.js to build a preview scene for the storage of core sketch histograms. The preview scene is rendered as a user-visible area for the display of core splicing and cataloging results, which can be dragged and scaled. The system creates a plane rectangular object in the preview scene and pastes the processed core grid onto it through the texture UV coordinates. The lower left corner of the image is the coordinate origin (0,0), and the upper right corner is the coordinate (1.0,1.0). The vertical and horizontal coordinates of all positions on the image are between 0.0 and 1.0, ensuring that each cropped core grid is completely mapped to the plane rectangular object. According to the coordinate information carried by the plane rectangular object, each single core grid can be spliced ​​by depth to form a drilling core sketch histogram displayed by depth.

[0092] For another example, the system also uses the 3D scene library three.js to create a printing scene in the system background for rendering graphic content. The printing scene uses the configuration option buffer (preserveDrawingBuffer) in the renderer (WebGLRenderer) to generate a graphics buffer. The printing scene uses the line object Line in the 3D scene to create a table and grid object, draws the table according to the aspect ratio of A4 paper and the table content style, and adds text annotations through css2d (a tool for setting annotations and prompt information in the printing scene). The text is marked at a fixed position according to the depth of the drilling catalog results and is scaled proportionally.

[0093] Finally, the system print scene will render the content of each page of the preview scene to the canvas in a loop according to the height of the A4 paper and the total length of the borehole. The rendered content includes the table header, grid lines, structured data annotations, and the core sketch of the current page spliced ​​in depth order. After the rendering is completed, the canvas generated for each page will be converted to a PDF file of A4 paper size for merging. The merging rule is to have a separate PDF file for 200 pages. Finally, the structured data in the system together with the core sketch of the borehole will be Figure 1 Upload to the system database for storage, preview and printing.

[0094] The above method of quickly generating sketches based on real-life photos of cores can intelligently detect the angle and range of the required core cutting through the core photos taken at the construction site by introducing a mature artificial intelligence graphic recognition algorithm. After setting the cutting parameters of the core photos, the core photos are cut into single core grids in equal proportions, and automatically spliced ​​according to the depth to form a sketch of the drilled core. Then, the drilling geological information is collected through the structured data cataloging form in the system, and the cataloging results are directly marked on the sketch of the drilled core according to the depth, including a detailed description of the geological structure of the core, in-situ test data, sampling information, and structured data such as groundwater level. The electronic collection method also stipulates a unified standard for the placement of cores and taking photos, supports the field exploration operation mode of paperless office, abandons the errors and low work efficiency caused by manual hand-drawing, improves the efficiency of field work, and can also play a role in field supervision through big data traces, and prevents the falsification of drilling cataloging results.

[0095] The following is a description of the device for quickly generating sketches based on real core photos provided by the present invention. The device for quickly generating sketches based on real core photos described below and the method for quickly generating sketches based on real core photos described above can be referenced to each other.

[0096] like Fig. 9 As shown, in one embodiment, a device for quickly generating a sketch based on a real-life photo of a core includes a core range detection module 910 , a core grid cutting module 920 , a core grid splicing module 930 , and a visualization display module 940 .

[0097] The core range detection module 910 is used to obtain a core box photo and rotate the core box photo to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the corresponding core grid in the core box photo to obtain the core range.

[0098] The core grid cutting module 920 is used to cut the core grid based on the input cutting boundary parameters, obtain the core cutting parameters, and attach the core box bottom depth value corresponding to the core cutting parameters to each cut core grid.

[0099] The core grid splicing module 930 is used to directly splice the core grids whose core recovery rate exceeds the first threshold value into a sketch diagram according to the well depth or footage, and to stretch the local core missing sections of the core grids whose core recovery rate is lower than the first threshold value and then splice them into a sketch diagram to obtain a core sketch column diagram.

[0100] The visualization display module 940 is used to construct a preview scene for the storage of the core sketch bar chart using a 3D scene library, and the preview scene is used to visualize the sketch chart formed by splicing the core grid and the cataloging results.

[0101] The core cutting parameters include the core box length, the number of core grids and the core box bottom depth.

[0102] In this embodiment, the present invention provides a device for quickly generating a sketch based on a real-life photo of a core, and the core range detection module is specifically used for:

[0103] The core box photo is captured, and after the core box photo is captured, the Matrix operation matrix of the picture is set to obtain the core box picture matrix.

[0104] By setting the x and y directions in the core box image matrix, the core box photo is scaled at the set scaling ratio.

[0105] In this embodiment, the device for quickly generating a sketch based on a real-life photo of a core provided by the present invention, the core range detection module is specifically further used for:

[0106] Based on the scaled core box photo, the rotation angle of the core box photo is detected, and ExifInterface is called to obtain the rotation angle of the core box photo according to the information carried by the core box photo.

[0107] According to the rotation angle of the core box photo, the core box photo is rotated to the normal viewing angle with the axis of the x and y axes as the reference point.

[0108] In this embodiment, the device for quickly generating a sketch based on a real-life photo of a core provided by the present invention, the core range detection module is specifically further used for:

[0109] The core range in the core box photo is detected by the single-class object instance segmentation method in the Mask-RCNN instance segmentation algorithm, and the Mask R-CNN is set for the pixel points of the core object within the core range.

[0110] The Mask-RCNN instance segmentation model is trained using multiple core box photos to determine the core range, and a pre-trained neural network is obtained.

[0111] In this embodiment, the device for quickly generating a sketch based on a real-life photo of a core provided by the present invention, the core range detection module is specifically further used for:

[0112] The core box photo to be detected is used as the input of the pre-trained neural network, so as to call the pre-trained neural network to perform convolution and pooling processing on the pixel points of the core box photo to obtain the prediction result of the core range.

[0113] The prediction result is compared with the core box photo to be detected to obtain the alignment deviation value between the pixels of the prediction result and the core box photo to be detected, and the mask set for the core pixel points is calculated by bilinear interpolation to make the alignment deviation value reach the second threshold.

[0114] In this embodiment, the present invention provides a device for quickly generating a sketch based on a real-life photo of a core, and the core grid splicing module is specifically used for:

[0115] It is detected whether there is a local core missing section at the break of the taken core, and when there is a local core missing section at the break of the taken core, the missing section of the single core is longitudinally stretched according to the preset depth to make the single core reach the actual depth position.

[0116] The starting and ending positions of the missing section of a single core are marked to determine the core interval to be stretched, and the core interval is longitudinally stretched according to the actual depth so that each core reaches the actual depth position.

[0117] In this embodiment, the device for quickly generating a sketch based on a core real-life photo provided by the present invention further includes a rendering module for:

[0118] The 3D scene library is used to create a printing scene for rendering graphic content. The printing scene uses the configuration options in the renderer WebGLRenderer to generate a graphics buffer.

[0119] By printing the scene, the rendering content of each page of the preview scene is rendered to the canvas according to the printing paper height and the drilling length. The rendering content includes the header, grid lines, structured data annotations, and a sketch formed by splicing the current page in the order of the core box bottom depth values.

[0120] Fig.10 An example of a physical structure diagram of an electronic device is shown. The electronic device may be a smart terminal, and its internal structure diagram may be as follows: Fig.10 As shown. The electronic device includes a processor, a memory and a network interface connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for quickly generating a sketch based on a real-life photograph of a core is implemented, and the method includes:

[0121] Obtain a photo of the core box, and rotate the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the corresponding core grid in the photo of the core box to obtain the core range;

[0122] The core grid is cut based on the input cutting boundary parameters, and the core cutting parameters are obtained, and the core box bottom depth value corresponding to the core cutting parameters is attached to each cut core grid;

[0123] According to the well depth or footage, the core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​into a sketch map to obtain a core sketch column map;

[0124] A preview scene is constructed using the 3D scene library for the storage of core sketch histograms. The preview scene is used to visualize the sketches formed by the splicing of core grids and the cataloging results.

[0125] The core cutting parameters include the core box length, the number of core grids and the core box bottom depth.

[0126] Those skilled in the art will understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present invention, and does not constitute a limitation on the electronic device to which the scheme of the present invention is applied. The specific electronic device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0127] On the other hand, the present invention also provides a computer storage medium storing a computer program, which, when executed by a processor, implements a method for quickly generating a sketch based on a core real scene photo, the method comprising:

[0128] Obtain a photo of the core box, and rotate the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the corresponding core grid in the photo of the core box to obtain the core range;

[0129] The core grid is cut based on the input cutting boundary parameters, and the core cutting parameters are obtained, and the core box bottom depth value corresponding to the core cutting parameters is attached to each cut core grid;

[0130] According to the well depth or footage, the core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​into a sketch map to obtain a core sketch column map;

[0131] A preview scene is constructed using the 3D scene library for the storage of core sketch histograms. The preview scene is used to visualize the sketches formed by the splicing of core grids and the cataloging results.

[0132] The core cutting parameters include the core box length, the number of core grids and the core box bottom depth.

[0133] In another aspect, a computer program product or a computer program is provided, the computer program product or the computer program comprising computer instructions, the computer instructions being stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium, and when the processor executes the computer instructions, a method for quickly generating a sketch based on a real-life photograph of a core is implemented, the method comprising:

[0134] Obtain a photo of the core box, and rotate the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the corresponding core grid in the photo of the core box to obtain the core range;

[0135] The core grid is cut based on the input cutting boundary parameters, and the core cutting parameters are obtained, and the core box bottom depth value corresponding to the core cutting parameters is attached to each cut core grid;

[0136] According to the well depth or footage, the core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​into a sketch map to obtain a core sketch column map;

[0137] A preview scene is constructed using the 3D scene library for the storage of core sketch histograms. The preview scene is used to visualize the sketches formed by the splicing of core grids and the cataloging results.

[0138] The core cutting parameters include the core box length, the number of core grids and the core box bottom depth.

[0139] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing related hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention may 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.

[0140] By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0141] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0142] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for quickly generating sketches based on real-life photos of rock cores, characterized in that: The method comprises: Obtain a photo of the core box, and rotate the photo of the core box to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the photo of the core box to obtain the core range; The core grid is trimmed based on the input trimming boundary parameters, and the core trimming parameters are obtained, and the core box bottom depth value corresponding to the core trimming parameters is attached to each of the trimmed core grids; According to the well depth or footage, the core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​into the sketch map to obtain a core sketch column map; A preview scene is constructed for the storage of the core sketch histogram using a 3D scene library, and the preview scene is used to visualize the sketch formed by splicing the core grid and the cataloging results; The core cutting parameters include the core box length, the number of core grids and the core box bottom depth. The method of obtaining a core box photo and rotating the core box photo to a front view angle by using an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range also includes: The core range in the core box photo is detected by using a single-class object instance segmentation method in a Mask-RCNN instance segmentation algorithm, and a Mask R-CNN is set for pixel points of the core object within the core range; The Mask-RCNN instance segmentation model is trained by using a plurality of different core box photos for determining the core range to obtain a pre-trained neural network; The method of obtaining a core box photo and rotating the core box photo to a front view angle by using an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range also includes: Using the core box photo to be detected as the input of the pre-trained neural network, so as to call the pre-trained neural network to perform convolution and pooling processing on the pixel points of the core box photo, and obtain the prediction result of the core range; The prediction result is compared with the core box photo to be detected to obtain an alignment deviation value between the pixels of the prediction result and the core box photo to be detected, and a mask for setting the core pixel points is calculated by bilinear interpolation so that the alignment deviation value reaches a second threshold.

2. The method for quickly generating a sketch based on a core real-life photograph according to claim 1, characterized in that: The core box photo is obtained, and the core box photo is rotated to a front view angle by an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range, including: Capturing the core box photo, and setting a Matrix operation matrix of the picture after capturing the core box photo, to obtain a core box picture matrix; The core box photo is scaled by setting the x and y directions in the core box photo matrix and the scale ratio is set.

3. The method for quickly generating a sketch based on a core real scene photograph according to claim 2, characterized in that: The method of obtaining a core box photo and rotating the core box photo to a front view angle by using an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range also includes: Based on the scaled core box photo, detecting the rotation angle of the core box photo, and calling ExifInterface to obtain the rotation angle of the core box photo according to the information carried by the core box photo; According to the rotation angle of the core box photograph, the core box photograph is rotated to a normal viewing angle with the axis of the x-axis and the y-axis as a reference point.

4. The method for quickly generating a sketch based on a core real-life photograph according to claim 1, characterized in that: The core grids whose core recovery rate exceeds the first threshold are directly spliced ​​to form a sketch map according to the well depth or footage, and the local core missing sections of the core grids whose core recovery rate is lower than the first threshold are stretched and spliced ​​to form the sketch map, to obtain a core sketch column map, including: Detecting whether there is a local core missing section at the break of the taken core, and if there is a local core missing section at the break of the taken core, longitudinally stretching the missing section of the single core according to a preset depth so that the single core reaches the actual depth position; The starting position and the ending position of the missing section of the single core are marked to determine the core interval to be stretched, and the core interval is longitudinally stretched according to the actual depth so that each core reaches the actual depth position.

5. The method for quickly generating a sketch based on a core real-life photograph according to any one of claims 1 to 4, characterized in that: The method further comprises: The 3D scene library is used to create a printing scene for rendering graphic content, wherein the printing scene uses configuration options in a renderer WebGLRenderer to generate a graphics buffer; Through the printing scene, each page of the preview scene is rendered to the canvas according to the printing paper height and the drilling length, and the rendering content includes a table header, grid lines, structured data annotations, and a sketch formed by splicing the current page in the order of the core box bottom depth values.

6. A device for quickly generating sketches based on real-life photos of rock cores, characterized in that: The device comprises: The core range detection module is used to obtain a core box photo and rotate the core box photo to a front view angle through an artificial intelligence graphic recognition algorithm to detect the range of the core grid corresponding to the core box photo to obtain the core range; A core grid cutting module is used to cut the core grid based on the input cutting boundary parameters, obtain the core cutting parameters, and attach the core box bottom depth value corresponding to the core cutting parameters to each of the cut core grids; A core grid splicing module is used to directly splice the core grids whose core recovery rate exceeds the first threshold value according to the well depth or footage to form a sketch map, and to stretch the local core missing sections of the core grids whose core recovery rate is lower than the first threshold value and then splice them into the sketch map to obtain a core sketch column map; A visualization display module, for constructing a preview scene for the storage of the core sketch histogram using a 3D scene library, wherein the preview scene is used for visually displaying the sketch formed by splicing the core grid and the cataloging results; The core cutting parameters include the core box length, the number of core grids and the core box bottom depth. The core range detection module is also specifically used for: The core range in the core box photo is detected by using the single-class object instance segmentation method in the Mask-RCNN instance segmentation algorithm, and the mask R-CNN is set for the pixel points of the core object within the core range; The Mask-RCNN instance segmentation model is trained using multiple core box photos to determine the core range, obtaining a pre-trained neural network. The core range detection module is also specifically used for: The core box photo to be detected is used as the input of the pre-trained neural network, so as to call the pre-trained neural network to perform convolution and pooling processing on the pixel points of the core box photo to obtain the prediction result of the core range; The prediction result is compared with the core box photo to be tested to obtain the alignment deviation value between the pixels of the prediction result and the core box photo to be tested, and the mask set for the core pixel points is calculated by bilinear interpolation to make the alignment deviation value reach the second threshold.

Images