Rock core quality index measuring and calculating method, system and equipment and storage medium

Abstract

The invention relates to a rock core quality index measuring and calculating method, system and equipment and a storage medium, and mainly relates to a method for measuring and calculating a rock core quality index, the side surfaces and the top and bottom surfaces of rock core sections in a rock core image are identified by using a trained semantic segmentation model, whether every two adjacent rock core sections meet mechanical fracture conditions is judged based on similarity indexes, and if yes, the rock core quality index is calculated. If yes, combining a plurality of mechanically fractured rock core sections as merged rock core sections, obtaining the lengths of the rock core sections by utilizing the side surfaces of the rock core sections, then obtaining the rock core sections of which the lengths meet the threshold requirement and do not meet the mechanical fracture condition and the merged rock core sections meeting the mechanical fracture condition, and calculating the quality of the rock core according to the lengths and a rock core quality index calculation formula. Therefore, the rock core quality index is obtained, and the accuracy of rock core quality index calculation is realized.

Description

Technical Field

[0001] The present application relates to the technical field of engineering geological exploration, and in particular to a method, system, equipment and storage medium for calculating core quality indicators. Background Art

[0002] In engineering geological exploration, core collection and analysis are important steps in assessing rock mass stability. Rock Quality Designation (RQD) is an important parameter for assessing rock mass integrity, which is usually determined by calculating the proportion of core sections with a length greater than 10 cm in the total coring length.

[0003] The existing RQD calculation method fails to take into account the unnatural fracture of the core caused by mechanical action, which usually leads to smaller measurement results and low data accuracy. Summary of the invention

[0004] The present application proposes a core quality index measurement method, system, device and storage medium to improve the RQD measurement accuracy.

[0005] In a first aspect, a method for calculating a core quality index is provided, comprising:

[0006] Obtaining the current core image after preprocessing;

[0007] Using the trained semantic segmentation model, the current core image is processed to identify the side surfaces and top and bottom surfaces of several core segments contained in the current core image;

[0008] Based on the similarity judgment of the top and bottom surfaces of two adjacent core segments, the core segments that meet the mechanical fracture conditions are merged as the merged core segments;

[0009] The length of the core section is obtained from the side;

[0010] Selecting a first batch of core segments whose lengths meet the preset threshold requirements from the core segments that do not meet the mechanical fracture conditions, and selecting a second batch of core segments whose lengths after merging meet the preset threshold requirements from the merged core segments;

[0011] Calculate the sum of the lengths of the first batch of core segments and the second batch of core segments;

[0012] The length and are used as parameters of a core quality index calculation formula to obtain a core quality index.

[0013] In certain embodiments of the first aspect, the pretreatment comprises:

[0014] Obtaining the endpoint coordinates of the core box in the length and width directions in the current core image;

[0015] Set the image size after perspective transformation, which fills the core box in length and width directions. The length and width ratio of the image is the same as that of the core box.

[0016] According to the image size after perspective transformation and the size of the core box, the conversion scale between the image and the actual size is calculated;

[0017] Calculating perspective transformation parameters according to the endpoint coordinates and the image size;

[0018] The perspective transformation parameters are used to perform perspective transformation on the core image.

[0019] In certain implementations of the first aspect, the preprocessing further includes: format conversion, brightness adjustment, and filtering and denoising.

[0020] In certain embodiments of the first aspect, the method further comprises:

[0021] Obtaining preprocessed basic core images for training;

[0022] Using image flipping technology, the basic core image for training is expanded to form a training image set;

[0023] Dividing the training image set into a training set and a validation set;

[0024] The training set is trained using the initial model, and the training result is evaluated using the validation set to obtain the semantic segmentation model.

[0025] In certain embodiments of the first aspect, obtaining the length of the core segment from the side specifically includes:

[0026] Using image processing techniques, the side profile of the core section is extracted;

[0027] According to the side profile, calculating a virtual length value of the core segment in the core image after perspective transformation;

[0028] The actual length of the core section is obtained according to the virtual length value and the conversion scale.

[0029] In certain embodiments of the first aspect, core segments that meet the mechanical fracture conditions are merged as merged core segments based on similarity judgment of top and bottom surfaces of two adjacent core segments, specifically including:

[0030] Use edge detection algorithm to obtain the top and bottom edges of the core section;

[0031] Establishing a coordinate system according to the side profile of the core section, and projecting the top and bottom surface edges into the coordinate system;

[0032] When the similarity of the top surface edge projections and the bottom surface edge projections corresponding to two adjacent core segments meets the first threshold requirement, extracting the top surface partial edge projections and the bottom surface partial edge projections whose similarity meets the second threshold requirement from the top surface edge projections and the bottom surface edge projections that meet the first threshold requirement;

[0033] Fitting the edge projection of the top surface portion and the edge projection of the bottom surface portion into an ellipse and a line segment;

[0034] Calculate the correlation coefficients of ellipses and line segments respectively;

[0035] When the correlation coefficient meets the third threshold requirement, the two adjacent core segments meet the mechanical fracture condition.

[0036] In certain embodiments of the first aspect, the semantic segmentation model is a SegNeXt semantic segmentation model.

[0037] In the second aspect, a core quality index measurement system is provided, comprising:

[0038] A core image semantic segmentation unit is used to obtain a preprocessed current core image; process the current core image using a trained semantic segmentation model to identify the side surfaces and top and bottom surfaces of several core segments contained in the current core image; and,

[0039] The core quality index calculation unit is used to merge the core segments that meet the mechanical fracture conditions as the merged core segments based on the similarity judgment of the top and bottom surfaces of two adjacent core segments; obtain the length of the core segment from the side; select the first batch of core segments whose lengths meet the preset threshold requirements from the core segments that do not meet the mechanical fracture conditions, and select the second batch of core segments whose merged lengths meet the preset threshold requirements from the merged core segments; calculate the sum of the lengths of the first batch of core segments and the second batch of core segments; use the sum of the lengths as a parameter of the core quality index calculation formula to obtain the core quality index.

[0040] In a third aspect, a computer device is provided, comprising: a processor, configured to execute computer instructions stored in a memory, so that the computer device executes the method of the first aspect.

[0041] In a fourth aspect, a computer-readable storage medium is provided, on which computer instructions are stored. When the computer instructions are run on a computer, the method of the first aspect is executed.

[0042] The present application mainly involves using a trained semantic segmentation model to identify the side and top and bottom surfaces of the core segment in the core image when calculating the core quality index, and then judging whether the two adjacent core segments meet the mechanical fracture conditions based on the similarity of the top and bottom surfaces. If so, several mechanically fractured core segments are merged as merged core segments, and their lengths are obtained using the side surfaces of the core segments. Then, the core segments whose lengths meet the threshold requirements and do not meet the mechanical fracture conditions, and the merged core segments whose lengths meet the threshold requirements and meet the mechanical fracture conditions can be obtained. The core quality index is calculated by their lengths and the merged core segments. The core quality index is obtained by the formula, which not only realizes the accurate identification of the side and top and bottom surfaces of the core segment through the semantic segmentation model, but also obtains the accurate core segment length data and the top and bottom surface data of two adjacent core segments. The accuracy of these data ensures the accuracy of the subsequent data calculation of the core quality index. Moreover, the similarity of the top and bottom surface data is used to accurately identify the unnatural fracture of the core caused by mechanical action, and the mechanical fracture and the core segment length value are fully considered in the parameters required for the core quality index calculation formula, so that the core quality index calculation is more accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of a method for calculating core quality indicators of an embodiment of the present application;

[0044] Figure 2 It is a specific example flow chart of the method for calculating the core quality index of the present application;

[0045] Figure 3 It is a schematic diagram of the perspective transformation of the core image in the embodiment of the present application. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0047] The technical solution of the present application is described below in conjunction with the accompanying drawings.

[0048] like Figure 1 As shown, the present application proposes a method for calculating core quality indicators, including:

[0049] S101, obtaining a preprocessed current core image;

[0050] Specifically, the core is photographed by a camera to obtain an initial core image, and the initial core image is preprocessed to obtain the current core image;

[0051] The above-mentioned preprocessing may include perspective transformation, format conversion, brightness adjustment, filtering and denoising, etc.;

[0052] An optional implementation is that the core box is used to load the core. In the perspective transformation, firstly, the end point coordinates of the core box in the length and width directions in the current core image are obtained, that is, the end point coordinates of the four corners of the core box are usually obtained. Then, the image size after the perspective transformation and the core box is filled in the length and width directions is set. The length-to-width ratio of the image is the same as that of the core box. According to the image size after the perspective transformation and the size of the core box, the conversion scale between the image and the actual size is calculated. Then, according to the end point coordinates of the core box and the image size filled in the core box after the perspective transformation, the perspective transformation parameters are calculated. The core image can be perspective transformed by using the perspective transformation parameters.

[0053] The above conversion scale can be used for subsequent calculation of core section length;

[0054] S102, using the trained semantic segmentation model, processing the current core image to identify the side surfaces and top and bottom surfaces of several core segments contained in the current core image;

[0055] Specifically, the semantic segmentation model can assign a semantic label to each pixel in the image to indicate the category to which the pixel belongs. By using the trained semantic segmentation model that can identify pixels belonging to the side surface or top and bottom surfaces of the core from the core image, the current core image can be processed to identify pixels belonging to the side surface of the current core or pixels belonging to the top and bottom surfaces of the current core in the current core image.

[0056] An optional implementation is that, when training a semantic segmentation model, first obtain a number of pre-processed basic core images for training, and use image flipping technology to flip these basic core images for training at different angles, so as to expand the training samples and form a training image set. Then, the training image set is divided into a training set and a validation set according to a preset ratio. The training set can be trained using the initialized initial model, and the training results can be evaluated using the validation set. After the training is completed, the desired semantic segmentation model can be obtained.

[0057] The semantic segmentation model can be SegNeXt semantic segmentation model, or other semantic segmentation models, such as U-Net, etc.

[0058] S103, based on the similarity judgment of the top and bottom surfaces of two adjacent core segments, the core segments that meet the mechanical fracture conditions are merged as the merged core segments;

[0059] An optional implementation method is to first use an edge detection algorithm to obtain the top and bottom surface edges of the core segment, establish a coordinate system based on the side profile of the core segment extracted by image processing technology, and project the top and bottom surface edges into the coordinate system. When the similarity of the top surface edge projection and the bottom surface edge projection corresponding to two adjacent core segments meets the first threshold requirement (usually greater than or equal to the first threshold), extract the top surface partial edge projection and the bottom surface partial edge projection whose similarity meets the second threshold requirement (usually greater than or equal to the second threshold) from the top surface edge projection and the bottom surface edge projection that meet the first threshold requirement. Then, fit the extracted top surface partial edge projection and the bottom surface partial edge projection into an ellipse and a line segment, and calculate the correlation coefficient of the ellipse and the line segment respectively. When the correlation coefficient meets the third threshold requirement (usually less than or equal to the third threshold), the two adjacent core segments meet the mechanical fracture condition.

[0060] Since the fractures of the core segments caused by joints and fissures generally appear as ellipses or straight lines when projected on the top and bottom edges (projections in the same coordinate system under different postures), the correlation coefficient is greater than the third threshold, indicating that the projections of the top and bottom edges of two adjacent core segments correspond to each other. At this time, the two adjacent core segments are not mechanically fractured, otherwise, they are mechanically fractured.

[0061] In this way, if the core is mechanically fractured, at least two core segments that meet the mechanical fracture conditions will be formed. Through the processing of this step, at least two core segments corresponding to the mechanical fracture of the core can be identified, and these core segments will be merged as the merged core segment;

[0062] S104, obtain the length of the core section from the side;

[0063] Specifically, the image processing technology can be used to extract the side profile of the core segment, and based on the side profile, the virtual length value of the core segment in the core image after perspective transformation can be calculated, and then, based on the virtual length value and the conversion scale used in the above perspective transformation, the actual length of the core segment can be obtained;

[0064] S105, selecting a first batch of core segments whose lengths meet a preset threshold requirement from the core segments that do not meet the mechanical fracture condition, and selecting a second batch of core segments whose lengths after merging meet the preset threshold requirement from the merged core segments;

[0065] Specifically, through the above similarity judgment, it can be determined which core segments are caused by mechanical fracture of the core, and these core segments are the merged core segments. In this way, the first batch of core segments whose lengths meet the preset threshold requirements can be selected from other core segments that have not been merged and are not merged core segments, and the second batch of core segments whose merged lengths meet the preset threshold requirements can also be selected from the merged merged core segments;

[0066] S106, calculating and obtaining the sum of the lengths of the first batch of core segments and the second batch of core segments;

[0067] S107, using the length and as parameters of a core quality index calculation formula to obtain a core quality index;

[0068] Specifically, the core quality indicator RQD may be obtained by using the sum of the lengths of the first batch of core segments and the second batch of core segments and the ratio of the total coring length.

[0069] Accordingly, this application may also involve the following products during its specific implementation:

[0070] A core quality index measurement system, comprising:

[0071] A core image semantic segmentation unit is used to obtain a preprocessed current core image; process the current core image using a trained semantic segmentation model to identify the side surfaces and top and bottom surfaces of several core segments contained in the current core image; and,

[0072] The core quality index calculation unit is used to merge the core segments that meet the mechanical fracture conditions as the merged core segments based on the similarity judgment of the top and bottom surfaces of two adjacent core segments; obtain the length of the core segment from the side; select the first batch of core segments whose lengths meet the preset threshold requirements from the core segments that do not meet the mechanical fracture conditions, and select the second batch of core segments whose merged lengths meet the preset threshold requirements from the merged core segments; calculate the sum of the lengths of the first batch of core segments and the second batch of core segments; use the sum of the lengths as a parameter of the core quality index calculation formula to obtain the core quality index.

[0073] A computer device comprises: a processor for executing computer instructions stored in a memory, so that the computer device executes the above-mentioned method for measuring core quality indicators.

[0074] A computer-readable storage medium stores computer instructions. When the computer instructions are run on a computer, the method for calculating the core quality index is executed.

[0075] In this way, not only can the semantic segmentation model be used to accurately identify the side surfaces and top and bottom surfaces of the core segment, and then obtain accurate core segment length data and top and bottom surface data of two adjacent core segments, the accuracy of these data ensures the accuracy of the subsequent data calculation of core quality indicators. Moreover, by utilizing the similarity of the top and bottom surface data, the unnatural fractures of the core caused by mechanical action can be accurately identified, and the mechanical fractures and the core segment length values ​​are fully considered in the parameters required for the core quality indicator calculation formula, thereby making the core quality indicator calculation more accurate.

[0076] The following describes the method for calculating the core quality index of the present application through a specific example.

[0077] like Figure 2 As shown, the core quality index calculation method of this specific example includes: a core image preprocessing process, a core image semantic segmentation process and a core RQD calculation process. The core image preprocessing process is used to preprocess the core image to facilitate segmentation. The core image semantic segmentation process is used to divide the pixels in the preprocessed core image into three categories: core side, core top and bottom surface and background. The core RQD calculation process is used to identify the core size and calculate the core RQD.

[0078] The core image preprocessing process includes the following steps:

[0079] Step 1.1.1: Obtain the coordinates of the four corners of the core box in the core image;

[0080] Step 1.1.2: Set the image size after perspective transformation, which fills the core box in length and width directions. The length and width ratio of the image is the same as that of the core box. Calculate the conversion scale between the image and the actual size based on the image size after perspective transformation and the core box size. For a specific example of this step, refer to Figure 3 ;

[0081] Step 1.1.3: Calculate the perspective transformation parameters based on the coordinates of the four corners of the core box in the core image and the image size filled with the core box, and then perform perspective transformation;

[0082] Step 1.2.1: Convert the perspective transformed core image format from RGB format to HSV format;

[0083] Step 1.2.2: Perform adaptive histogram equalization on the brightness channel to adjust the brightness of the core image;

[0084] Step 1.2.3: Convert the core image format after brightness adjustment from HSV format to RGB format;

[0085] Step 1.3: Perform bilateral filtering to denoise the core image.

[0086] The core image semantic segmentation process uses the trained SegNext model to perform semantic segmentation on the core image, thereby dividing the pixels in the core image into three categories: core side, core top and bottom, and background.

[0087] The training of the SegNext model consists of the following steps:

[0088] Step 2.1: using at least 500 pre-processed core images obtained by the core image pre-processing process;

[0089] Step 2.2: Use annotation software or image processing technology to assist in labeling the pre-processed core image;

[0090] Step 2.3: Expand the marked core image by 90° rotation, 180° rotation, 270° rotation and horizontal flipping to form a core image dataset;

[0091] Step 2.4: Divide the core image dataset into a training set and a validation set in a ratio of 8:2;

[0092] Step 2.5: Set the training parameters of the SegNext model, which mainly include learning rate, training cycle and batch size;

[0093] Step 2.6: Start cyclic training of SegNeXt using the training set. After each cycle, use the validation set to evaluate the model performance.

[0094] Step 2.7: After training, the optimal SegNeXt core segmentation model is obtained based on the evaluation results of the validation set.

[0095] The core RQD calculation process includes the following steps:

[0096] Step 3.1: Extract the side profile of the core section using image processing technology;

[0097] Step 3.2: Calculate the length of the core segment using the side profile of the core segment, and calculate the actual length of the core segment based on the conversion scale between the image and the actual size;

[0098] Step 3.3: Determine the core fracture factors based on the top and bottom surfaces of the core section;

[0099] Step 3.4: Add the lengths of adjacent core segments due to mechanical fracture and group them into one core segment;

[0100] Step 3.5: Calculate the length and the length of the core segments ≥10 cm;

[0101] Step 3.6: Calculate the core RQD according to the core RQD calculation formula.

[0102] In the above step 3.3, determining the core fracture factors includes the following steps:

[0103] Step 4.1: Use edge detection algorithm to obtain the edges at both ends of each core segment;

[0104] Step 4.2: Establish a coordinate system based on the side profile of the core segment obtained in step 3.1, and project the edges of the two ends of the core segment into the coordinate system at a ratio of 1:1, so as to evaluate the similarity between the edge of the core segment and the edge of its adjacent core segment;

[0105] Step 4.3: If the similarity is >70%, extract the edge of the core segment with a similarity >85% and fit it into ellipses and line segments, otherwise it is considered that the two core segments are not mechanically fractured;

[0106] Step 4.4: Calculate the correlation coefficients of the ellipse and line segment respectively;

[0107] Step 4.5: If both correlation coefficients are <90%, it is considered that the two core sections are caused by mechanical fracture, otherwise it is considered that the two core sections are not caused by mechanical fracture.

[0108] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for measuring core quality indicators, characterized in that: include: Obtaining the preprocessed current core image; Using the trained semantic segmentation model, the current core image is processed to identify the side surfaces and top and bottom surfaces of several core segments contained in the current core image; Based on the similarity judgment of the top and bottom surfaces of two adjacent core segments, the core segments that meet the mechanical fracture conditions are merged as the merged core segments; The length of the core section is obtained from the side; Selecting a first batch of core segments whose lengths meet the preset threshold requirements from the core segments that do not meet the mechanical fracture conditions, and selecting a second batch of core segments whose lengths after merging meet the preset threshold requirements from the merged core segments; Calculate the sum of the lengths of the first batch of core segments and the second batch of core segments; The length and are used as parameters of a core quality index calculation formula to obtain a core quality index.

2. The method for calculating the core quality index according to claim 1, characterized in that: The pre-processing comprises: Obtaining the endpoint coordinates of the core box in the length and width directions in the current core image; Set the image size after perspective transformation, which fills the core box in length and width directions. The length and width ratio of the image is the same as that of the core box. According to the image size after perspective transformation and the size of the core box, the conversion scale between the image and the actual size is calculated; Calculating perspective transformation parameters according to the endpoint coordinates and the image size; The perspective transformation parameters are used to perform perspective transformation on the core image.

3. The method for calculating the core quality index according to claim 2, characterized in that: The preprocessing also includes: format conversion, brightness adjustment and filtering and denoising.

4. The method for calculating the core quality index according to claim 1, characterized in that: The method further includes: obtaining a preprocessed basic core image for training; Using image flipping technology, the basic core image for training is expanded to form a training image set; the training image set is divided into a training set and a verification set; The training set is trained using the initial model, and the training result is evaluated using the validation set to obtain the semantic segmentation model.

5. The method for calculating the core quality index according to claim 2, characterized in that: The length of the core section obtained from the side specifically includes: Using image processing techniques, the side profile of the core section is extracted; According to the side profile, a virtual length value of the core segment in the core image after perspective transformation is calculated; according to the virtual length value and the conversion scale, the actual length of the core segment is obtained.

6. The method for calculating the core quality index according to claim 5, characterized in that: Based on the similarity judgment of the top and bottom surfaces of two adjacent core segments, the core segments that meet the mechanical fracture conditions are merged as the merged core segments, including: Use edge detection algorithm to obtain the top and bottom edges of the core section; Establishing a coordinate system according to the side profile of the core section, and projecting the top and bottom surface edges into the coordinate system; When the similarity of the top surface edge projections and the bottom surface edge projections corresponding to two adjacent core segments meets the first threshold requirement, extracting the top surface partial edge projections and the bottom surface partial edge projections whose similarity meets the second threshold requirement from the top surface edge projections and the bottom surface edge projections that meet the first threshold requirement; Fitting the edge projection of the top surface portion and the edge projection of the bottom surface portion into an ellipse and a line segment; Calculate the correlation coefficients of ellipses and line segments respectively; When the correlation coefficient meets the third threshold requirement, the two adjacent core segments meet the mechanical fracture condition.

7. The method for calculating the core quality index according to any one of claims 1 to 6, characterized in that: The semantic segmentation model is a SegNeXt semantic segmentation model.

8. A core quality index measurement system, characterized in that: include: A core image semantic segmentation unit is used to obtain a preprocessed current core image; use the trained semantic segmentation model to process the current core image and identify the side surfaces and top and bottom surfaces of several core segments contained in the current core image; and The core quality index calculation unit is used to merge the core segments that meet the mechanical fracture conditions as the merged core segments based on the similarity judgment of the top and bottom surfaces of two adjacent core segments; obtain the length of the core segment from the side; select the first batch of core segments whose lengths meet the preset threshold requirements from the core segments that do not meet the mechanical fracture conditions, and select the second batch of core segments whose merged lengths meet the preset threshold requirements from the merged core segments; calculate the sum of the lengths of the first batch of core segments and the second batch of core segments; use the sum of the lengths as a parameter of the core quality index calculation formula to obtain the core quality index.

9. A computer device, characterized in that: include: A processor, configured to execute computer instructions stored in the memory, so that the computer device executes the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: Computer instructions are stored thereon, and when the computer instructions are run on a computer, the method according to any one of claims 1 to 8 is executed.

Images