Visual estimation method of three-dimensional size of rockfill

Abstract

The present application belongs to the field of rockfill material gradation detection, and relates to a visual estimation method for three-dimensional size of rockfill material, comprising the following steps: 1) obtaining an image of mixed gradation rockfill material, and performing two-dimensional contour segmentation of rockfill material particles on the image of mixed gradation rockfill material; 2) extracting an estimated value of three-dimensional size of rockfill material particles based on the result obtained in step 1); and 3) estimating three-dimensional size of rockfill material based on the estimated value of three-dimensional size of rockfill material particles obtained in step 2). The present application provides a visual estimation method for three-dimensional size of rockfill material for gradation analysis, which has high identification accuracy and can effectively improve detection efficiency.

Description

Technical Field

[0001] This invention belongs to the field of riprap gradation detection, and relates to a visual estimation method for the three-dimensional dimensions of riprap, and more particularly to a visual estimation method for the three-dimensional dimensions of riprap for gradation analysis based on computer vision and deep learning technology. Background Technology

[0002] The gradation of aggregates (the percentage of aggregates with different particle sizes by weight in the total aggregate) is a key indicator for quality control in building material production. The gradation of rockfill is a crucial aspect of quality control in rockfill dam construction. Traditional methods for testing rockfill gradation primarily rely on sieve analysis, which typically requires significant manual labor. This method is not only inefficient and costly, but also yields limited representativeness of the results, making it difficult to meet the demands of rapid and precise construction in engineering projects.

[0003] With the rapid development of information technology, computer vision technology has been gradually introduced into the field of riprap gradation detection. Detection methods based on two-dimensional vision perception have become a research hotspot due to their high efficiency and low cost. These methods analyze two-dimensional images of riprap to infer particle size and gradation, reducing the limitations of traditional methods to some extent. However, the conversion from two-dimensional images of the stone to three-dimensional dimensions remains a key technical challenge. Current research commonly uses ellipse fitting technology to process the two-dimensional contour of the stone, calculate the equivalent particle size, and derive the particle volume and gradation using the ellipsoidal volume formula. While this has achieved initial success in improving detection efficiency and achieving automation, it has not effectively solved the problem of contour information loss caused by particle stacking and occlusion during the imaging process of riprap. Furthermore, it lacks adaptability to the irregular shapes of stone particles and their stacking posture in three-dimensional space, resulting in limited accuracy in gradation characterization.

[0004] Therefore, overcoming the influence of stacking angle and stacking occlusion on particle size estimation has become a core problem that urgently needs to be solved in the improvement of gradation detection technology for riprap based on two-dimensional images. Summary of the Invention

[0005] In order to solve the above-mentioned technical problems in the background art, the present invention provides a visual estimation method for the three-dimensional dimensions of stone for gradation analysis, which has high identification accuracy and can effectively improve detection efficiency.

[0006] To achieve the above objectives, the present invention adopts the following technical solution:

[0007] A visual estimation method for the three-dimensional dimensions of riprap, characterized in that: the visual estimation method for the three-dimensional dimensions of riprap includes the following steps:

[0008] 1) Obtain images of mixed-graded rockfill and perform two-dimensional contour segmentation of the rockfill particles in the images of mixed-graded rockfill;

[0009] 2) Based on the results obtained in step 1), extract the estimated three-dimensional dimensions of the riprap particles;

[0010] 3) Estimate the three-dimensional dimensions of the rockfill based on the estimated three-dimensional dimensions of the rockfill particles obtained in step 2).

[0011] Preferably, step 1) is implemented as follows:

[0012] 1.1) Collect image samples of mixed-graded riprap under different imaging conditions;

[0013] 1.2) The image samples obtained in step 1.1) are trained based on YOLOv8m-seg to obtain a two-dimensional contour segmentation model of the riprap particles;

[0014] 1.3) Based on the two-dimensional contour segmentation model of the riprap particles obtained in step 1.2), the two-dimensional contour of the riprap particles is segmented.

[0015] Preferably, step 2) is implemented as follows:

[0016] 2.1) Based on the two-dimensional contour of the rockfill particles obtained in step 1), the key points of the two-dimensional contour of the rockfill particles are extracted using the convex hull algorithm.

[0017] 2.2) Fit the result obtained in step 2.1) to an equivalent ellipse, approximate the particle as an ellipsoid, and the equivalent ellipse is the projection of the particle approximate ellipsoid onto the plane, and obtain the equation of the projection relationship.

[0018] 2.3) Solve the equation of the projection relationship obtained in step 2.2) to obtain the estimated value of the three-dimensional size of the riprap particles.

[0019] Preferably, step 2.2) is implemented as follows:

[0020] The equation of the ellipsoid is:

[0021]

[0022] The equation of the equivalent ellipse is:

[0023] a 11 X 2 +2a 12 XY+a 22 Y 2 +a 33 =0(2)

[0024] in:

[0025]

[0026] in:

[0027] θ and These are the azimuth and elevation angles between the projection plane and the principal plane of the ellipsoid, respectively.

[0028] a, b, and c are the three semi-major axes of the ellipsoid;

[0029] The equation for the projection relationship between the length A and the minor semi-axis B of the ellipse and the three semi-major axes a, b, and c of the ellipsoid, which has equation (2), is:

[0030]

[0031] in:

[0032] I = a 11 +a 22 ;

[0033]

[0034] Preferably, step 2.3) is implemented as follows:

[0035] Introducing particle shape characteristics such as aspect ratio λ1, thickness-to-width ratio λ2, as well as stacking azimuth angle θ and pitch angle. probability distribution:

[0036] λ1=b / a, λ1~P1; λ2=c / b, λ2~P2; θ~P3;

[0037] The Monte Carlo method is used to solve for the feasible set of three-dimensional particle sizes, and the expectation of the solution set is used as the representative solution. The conditions for a feasible solution are:

[0038]

[0039] in:

[0040] n is the particle number of the riprap;

[0041] ε A and ε B These are the error thresholds for the major and minor semi-axis, respectively;

[0042] A * It is the dimension of the semi-major axis of the equivalent ellipse;

[0043] B * It is the dimension of the minor semi-axis of the equivalent ellipse;

[0044] Based on the feasible solution space, the estimated values ​​of the lengths a, b, and c of the three semi-axis of the ellipse are obtained respectively; the expressions for the estimated values ​​of the lengths of the three semi-axis of the ellipse are as follows:

[0045]

[0046] Where N is the number of feasible solutions;

[0047] It is an estimated value of the semi-axis length 'a';

[0048] It is an estimated value of the semi-axis length b;

[0049] It is an estimated value of the semi-axis length c.

[0050] Preferably, in step 2.3), the particle shape characteristics, such as the width-to-length ratio λ1, the thickness-to-width ratio λ2, the stacking azimuth angle θ, and the pitch angle, are... The specific implementation of the probability distribution is as follows:

[0051] The steps to obtain the probability distributions of particle shape characteristic indices λ1 (width-to-length ratio) and λ2 (thickness-to-width ratio) are as follows:

[0052] Randomly collect rockfill samples, measure the size of rockfill particles within each particle size range, and obtain the width-to-length ratio λ1 and thickness-to-width ratio λ2 particle shape characteristic indices. Fit the probability distribution models P1 and P2 of the corresponding particle shape characteristic indices respectively.

[0053] Obtain the azimuth angle θ and elevation angle of the piled stone particles. The steps for probability distribution are as follows:

[0054] a.1) Prepare dispersed stone samples, take photos of the samples, and obtain the graded GIR through standard sieve analysis;

[0055] a.2) Use the two-dimensional contour segmentation model of the riprap particles to obtain the two-dimensional contour of the stone sample particles in step a.1), and extract the equivalent ellipse major and minor semi-axis length values ​​A. * and B * ;

[0056] a.3) Randomly initialize the azimuth angle θ and elevation angle probability distribution model parameters: θ~U(θ min ,θ max ), The upper and lower limits are both within the range of [0, π / 2]; U is uniformly distributed.

[0057] a.4) Based on the equivalent ellipse semi-axis length value A obtained in step a.2), * and B * Width-to-length ratio λ1~P1 and thickness-to-width ratio λ2~P2 and The stone sample prepared in step a.1) was subjected to gradation GIR. * calculate;

[0058] a.5) The graded GIR obtained in step a.4) * The absolute error of the graded GIR in step a.1) is used as the optimization objective. The probability distribution model parameters of azimuth θ and elevation angle are adjusted by the grid optimization method until the absolute error is as small as possible. The error is then considered to have converged, and the probability distribution of azimuth θ and elevation angle is obtained.

[0059] Preferably, in step a.4), the stone sample undergoes gradation GIR. * The calculation method is as follows:

[0060]

[0061] in:

[0062] M is the mass of each particle of the rockfill in the stone sample;

[0063] n represents the total number of rockfill particles within this particle size range;

[0064] N' represents the total number of riprap particles in (1);

[0065] The expression for M is:

[0066]

[0067] Preferably, step 3) is implemented as follows:

[0068] Calculate the density ρ of the rockfill, the mass M of each particle, and the gradation GIR of the rockfill. * :

[0069]

[0070]

[0071] in:

[0072] n represents the total number of rockfill particles within this particle size range;

[0073] N' represents the total number of riprap particles in (1).

[0074] The advantages of this invention are:

[0075] This invention provides a visual estimation method for the three-dimensional dimensions of rockfill, comprising the following steps: 1) acquiring an image of mixed-graded rockfill and performing two-dimensional contour segmentation of the rockfill particles in the image; 2) extracting estimated values ​​of the three-dimensional dimensions of the rockfill particles based on the results obtained in step 1); 3) estimating the three-dimensional dimensions of the rockfill based on the estimated values ​​of the three-dimensional dimensions of the rockfill particles obtained in step 2). The visual estimation method for the three-dimensional dimensions of rockfill provided by this invention, combining statistical analysis, deep learning, and Monte Carlo estimation, can efficiently and accurately estimate the three-dimensional dimensions of rockfill particles, improving the accuracy of gradation characterization, and is applicable to quality control of rockfill dam construction and other geotechnical engineering fields. By combining statistical analysis and deep learning models, this method can efficiently process large-scale rockfill samples, significantly improving the efficiency of gradation analysis; simultaneously, through optimal equivalent ellipse and ellipsoidal geometric models, it greatly improves the accuracy of particle three-dimensional dimension estimation. This method utilizes the Monte Carlo method to consider the randomness of imaging angle and particle morphology, adapting to complex rockfill conditions, and achieving statistical approximation of particle three-dimensional dimensions, providing accurate data for gradation analysis and engineering design. It particularly emphasizes the impact of imaging angle on recognition results, and improves the recognition accuracy of irregular particle morphology by adjusting the azimuth and elevation angle distribution. It can maintain high efficiency in recognition even under complex imaging conditions, and has the characteristics of high detection efficiency, strong adaptability and high accuracy. Attached Figure Description

[0076] Figure 1 This is a flowchart of the visual estimation method for the three-dimensional dimensions of stone provided by the present invention.

[0077] Figure 2 It is a projection relationship diagram of the particle ellipsoid and its outline ellipse. Detailed Implementation

[0078] See Figure 1 This invention provides a visual estimation method for the three-dimensional dimensions of stone, specifically including the following steps:

[0079] 1) Obtain images of mixed-graded rockfill and perform two-dimensional contour segmentation of the rockfill particles in the images, specifically:

[0080] 1.1) Collect image samples of mixed-graded riprap under different imaging conditions;

[0081] 1.2) Based on YOLOv8m-seg, the image samples obtained in step 1.1) are trained to obtain a two-dimensional contour segmentation model for riprap particles; a riprap instance segmentation model is established through a convolutional neural network. This invention uses the YOLOv8m-seg model for contour segmentation of riprap particles. Based on this, images of mixed-graded riprap under different imaging conditions are collected, and the Labelme annotation tool is used to annotate the image samples with closed polygon contours. To improve the model's generalization ability and robustness, the dataset is augmented, specifically including brightness transformation, contrast enhancement, random rotation, horizontal flipping, and HSV data augmentation techniques. The dataset is divided into training and validation sets at a ratio of 80% to 20%. Before training the deep learning model, the weights of the YOLOv8m-seg model pre-trained on the COCO dataset are loaded for transfer learning. During model training, the validation set is used to initially verify the training results, and the model parameters are optimized and adjusted based on the accuracy performance on the validation set.

[0082] 1.3) Based on the two-dimensional contour segmentation model of the riprap particles obtained in step 1.2), the two-dimensional contour of the riprap particles is segmented. This invention extracts the particle contour using the `findContours` function. Then, the `convexHull` function is used to calculate the convex hull of the contour to obtain key feature points. Simultaneously, to achieve the best equivalent ellipse fitting, the `fitEllipse` function is used. This function, based on the least squares principle, can calculate the optimal fitted ellipse for a given set of contour points. This ellipse approximates the overall shape of the contour with the highest accuracy, ensuring the accuracy of the fitting result. The center coordinates, rotation angle, and dimensions of the major and minor axes of the ellipse can be extracted from the parameters returned by the `fitEllipse` function.

[0083] 2) Based on the results obtained in step 1), extract the estimated three-dimensional dimensions of the riprap particles, specifically:

[0084] 2.1) Based on the two-dimensional contour of the rockfill particles obtained in step 1), the key points of the two-dimensional contour of the rockfill particles are extracted using the convex hull algorithm.

[0085] 2.2) Based on the result obtained in step 2.1), fit an equivalent ellipse, approximating the particle as an ellipsoid. The equivalent ellipse is the projection of the approximate ellipsoid onto the plane, such as... Figure 2 As shown, the equation for obtaining the projection relationship is as follows:

[0086] The specific implementation method of step 2.2) is as follows:

[0087] The equation of the ellipsoid is:

[0088]

[0089] The equation of the equivalent ellipse is:

[0090] a 11 X 2 +2a 12 XY+a 22 Y 2 +a 33 =0(2)

[0091] in:

[0092]

[0093] in:

[0094] θ and Let θ be the azimuth and elevation angles between the projection plane and the principal plane of the ellipsoid, respectively, where θ ∈ [0, π / 2].

[0095] a, b, and c are the three semi-major axes of the ellipsoid;

[0096] The equation for the projection relationship between the length A and the minor semi-axis B of the ellipse and the three semi-major axes a, b, and c of the ellipsoid, which has equation (2), is:

[0097]

[0098]

[0099] in:

[0100] I = a 11 +a 22 ;

[0101]

[0102] 2.3) Solve the equation of the projection relationship obtained in step 2.2) to obtain the estimated values ​​of the three-dimensional dimensions of the riprap particles. Specifically, step 2.3) is implemented as follows:

[0103] Research in the field of geotechnical engineering shows that boulders exhibit a clear directional distribution pattern, which can be represented by a random normal distribution or a uniform distribution. Based on the measured particle shape characteristic distribution model, it is assumed that the center azimuth and elevation angles of particle imaging are discretely uniformly distributed. Sampling is performed on the aforementioned aspect ratio, width-to-thickness ratio, center azimuth angle, and elevation angle. Therefore, when solving the projection relationship equation, particle shape characteristic indices such as the width-to-length ratio λ1, the thickness-to-width ratio λ2, and the stacking azimuth angle θ and elevation angle are introduced. probability distribution:

[0104] λ1=b / a, λ1~P1; λ2=c / b, λ2~P2; θ~P3;

[0105] Particle shape characteristics include aspect ratio λ1, thickness-to-width ratio λ2, and stacking azimuth angle θ and elevation angle. The probability distribution, width-to-length ratio λ1, and thickness-to-width ratio λ2 can be directly obtained through on-site sampling measurements. The method is as follows: Place the particles in a position that maximizes their horizontal projected area, and measure the length, width, and thickness of their smallest circumscribed cuboid as particle shape characteristic parameters; stacking azimuth angle θ and pitch angle... Since these cannot be directly measured, they will not be discussed for now; let's assume the stacked azimuth angle θ and elevation angle are... The probability distribution is known; for example, the particle shape characteristics used in this invention include the width-to-length ratio λ1, the thickness-to-width ratio λ2, and the stacking azimuth angle θ and pitch angle. The specific implementation of the probability distribution is as follows:

[0106] The steps to obtain the probability distributions of particle shape characteristic indices λ1 (width-to-length ratio) and λ2 (thickness-to-width ratio) are as follows:

[0107] Randomly collect rockfill samples, measure the size of rockfill particles within each particle size range, and obtain the width-to-length ratio λ1 and thickness-to-width ratio λ2 particle shape characteristic indices. Fit the probability distribution models P1 and P2 of the corresponding particle shape characteristic indices respectively.

[0108] Obtain the azimuth angle θ and elevation angle of the piled stone particles. The steps for probability distribution are as follows:

[0109] a.1) Prepare dispersed stone samples, take photos of the samples, and obtain the graded GIR through standard sieve analysis;

[0110] a.2) Use the two-dimensional contour segmentation model of the riprap particles to obtain the two-dimensional contour of the stone sample particles in step a.1), and extract the equivalent ellipse major and minor semi-axis length values ​​A. * and B * ;

[0111] a.3) Randomly initialize the azimuth angle θ and elevation angle probability distribution model parameters: θ~U(θ min ,θ max ), The upper and lower limits both take values ​​in the range [0, π / 2]; U is a uniform distribution.

[0112] a.4) Based on the equivalent ellipse semi-axis length value A obtained in step a.2), * and B * Width-to-length ratio λ1~P1 and thickness-to-width ratio λ2~P2 and The stone sample prepared in step a.1) was subjected to gradation GIR. *calculate;

[0113] Among them, the stone samples were graded using GIR. * The calculation method is as follows:

[0114]

[0115] in:

[0116] M is the mass of each particle of the rockfill in the stone sample;

[0117] n represents the total number of rockfill particles within this particle size range;

[0118] N' represents the total number of riprap particles in (1);

[0119] The expression for M is:

[0120]

[0121] a.5) The graded GIR obtained in step a.4) * The absolute error of the graded GIR in step a.1) is used as the optimization objective. The probability distribution model parameters of azimuth θ and elevation angle are adjusted by the grid optimization method until the absolute error is as small as possible. The error is then considered to have converged, and the probability distribution of azimuth θ and elevation angle is obtained.

[0122] It should be noted that research on building materials shows that the natural stacking of stones exhibits a clear directional distribution pattern, and a uniform distribution can be used to represent the stacking angle conditions. Therefore, the azimuth angle θ and pitch angle can be assumed. If the distribution follows a uniform distribution and its form is known, then it is only necessary to determine the parameters of the distribution model.

[0123] The Monte Carlo method is used to solve for the feasible set of three-dimensional particle sizes, and the expectation of the solution set is used as the representative solution. The conditions for a feasible solution are:

[0124]

[0125] in:

[0126] n is the particle number of the riprap;

[0127] ε A and ε B These are the error thresholds for the major and minor semi-axis, respectively;

[0128] A * It is the dimension of the semi-major axis of the equivalent ellipse;

[0129] B * It is the dimension of the minor semi-axis of the equivalent ellipse;

[0130] Among them, A* and B * The extraction method is to determine the conversion relationship between the pixel size and actual size of the piled stone during the shooting stage by fixing the shooting distance, so as to determine the actual size of the equivalent ellipse. The major and minor semi-axes of the ellipse can be returned by using the fitEllipse function of the OpenCV library.

[0131] Based on the corresponding probability distribution, the aspect ratio λ1, the thickness-to-width ratio λ2, and the stacking azimuth θ and elevation angle are considered. Random sampling is performed. At this time, the system of equations in (1) only has one unknown among a, b, and c. Let's assume it's a. Similarly, we set an initial value for a and calculate a set of A and B. The calculated A and B are compared with the measured A. * and B * The two are compared, and if the difference exceeds the error threshold, the value of 'a' is adjusted until it is satisfied (Newton's method). At this point, the value of 'a' that satisfies the error threshold is a feasible solution. Knowing 'a', we can further calculate 'b' and 'c'. Repeating the above steps multiple times can yield multiple sets of feasible solutions, which are called the feasible solution space.

[0132] Based on the feasible solution space, the estimated values ​​of the lengths a, b, and c of the three semi-axis of the ellipse are obtained respectively; the expressions for the estimated values ​​of the lengths of the three semi-axis of the ellipse are as follows:

[0133]

[0134] Where N is the number of feasible solutions;

[0135] It is an estimated value of the semi-axis length 'a';

[0136] It is an estimated value of the semi-axis length b;

[0137] It is an estimated value of the semi-axis length c.

[0138] 3) Estimate the three-dimensional dimensions of the rockfill based on the estimated values ​​of the three-dimensional dimensions of the rockfill particles obtained in step 2). Specifically, the density of the rockfill is ρ, and the mass M and gradation GIR of each rockfill particle are calculated. * :

[0139]

[0140] in:

[0141] n represents the total number of rockfill particles within this particle size range;

[0142] N' represents the total number of riprap particles in (1).

Claims

1. A visual estimation method for the three-dimensional dimensions of riprap, characterized in that: The visual estimation method for the three-dimensional dimensions of the riprap includes the following steps: 1) Obtain images of mixed-graded rockfill and perform two-dimensional contour segmentation of the rockfill particles in the images of mixed-graded rockfill; 2) Based on the results obtained in step 1), extract the estimated three-dimensional dimensions of the riprap particles, specifically: 2.1) Based on the two-dimensional contour of the rockfill particles obtained in step 1), the key points of the two-dimensional contour of the rockfill particles are extracted using the convex hull algorithm. 2.2) Fit the result obtained in step 2.1) to an equivalent ellipse, approximate the particle as an ellipsoid, and the equivalent ellipse is the projection of the particle approximate ellipsoid onto the plane, and obtain the equation of the projection relationship. 2.3) Solve the equations of the projection relationship obtained in step 2.2) to obtain estimated values ​​of the three-dimensional dimensions of the riprap particles, specifically: Introducing the particle shape characteristic index width-to-length ratio Aspect Ratio and stacked azimuth angle and pitch angle probability distribution: ， ； ， ； ； （5） The Monte Carlo method is used to solve for the feasible set of three-dimensional particle sizes, and the expectation of the solution set is used as the representative solution. The conditions for a feasible solution are: ， （6） in: Number the granular material in the rockfill; and These are the error thresholds for the major and minor semi-axis, respectively. It is the dimension of the semi-major axis of the equivalent ellipse; It is the dimension of the minor semi-axis of the equivalent ellipse; Based on the feasible solution space, the lengths of the three semi-axes of the ellipse are obtained respectively. , , The estimated values; the expressions for the estimated values ​​of the three semi-axis lengths of the ellipse are as follows: （7） in, The number of feasible solutions; It is the length of the semi-axis The estimated value; It is an estimated value of the semi-axis length b; It is an estimated value of the semi-axis length c; Among them, the particle shape characteristic index width-to-length ratio Aspect Ratio and stacked azimuth angle and pitch angle The specific implementation of the probability distribution is as follows: Obtain the particle shape characteristic index width-to-length ratio Aspect Ratio The steps for probability distribution are as follows: Randomly collect rockfill samples, measure the size of rockfill particles within each particle size range, and obtain the aspect ratio. and aspect ratio For each grain shape characteristic index, a probability distribution model is fitted to that index. and ; Obtain the azimuth angle of the piled stone particles and pitch angle The steps for probability distribution are as follows: a.1) Prepare dispersed stone samples, photograph and sample them, and obtain the gradation through standard sieve analysis. ; a.2) Use the two-dimensional contour segmentation model of the riprap particles to obtain the two-dimensional contour of the stone sample particles in step a.1), and extract the equivalent ellipse major and minor axis length values. and ; a.3) Randomly initialize azimuth angle And pitch angle probability distribution model parameters: , The upper and lower limit value ranges are both The U is uniformly distributed; a.4) Based on the equivalent ellipse semi-axis length values ​​obtained in step a.2). and Width-to-length ratio and aspect ratio as well as , Grading of the stone sample prepared in step a.1) calculate; a.5) The gradation obtained in step a.4) Compared with the gradation in step a.1) The absolute error is used as the optimization objective, and the azimuth angle is adjusted through a mesh optimization method. The probability distribution model parameters for pitch and elevation angles are calculated until the absolute error is minimized, at which point the error is considered to have converged, and the azimuth angle is obtained. The probability distribution of pitch angle; 3) Estimate the three-dimensional dimensions of the rockfill based on the estimated three-dimensional dimensions of the rockfill particles obtained in step 2).

2. The visual estimation method for the three-dimensional dimensions of riprap according to claim 1, characterized in that: The specific implementation method of step 1) is as follows: 1.1) Collect image samples of mixed-graded riprap under different imaging conditions; 1.2) The image samples obtained in step 1.1) are trained based on YOLOv8m-seg to obtain a two-dimensional contour segmentation model of the riprap particles; 1.3) Based on the two-dimensional contour segmentation model of the riprap particles obtained in step 1.2), the two-dimensional contour of the riprap particles is segmented.

3. The visual estimation method for the three-dimensional dimensions of riprap according to claim 2, characterized in that: The specific implementation method of step 2.2) is as follows: The equation of the ellipsoid is: （1） The equation of the equivalent ellipse is: （2） in: （3） in: and These are the azimuth and elevation angles between the projection plane and the principal plane of the ellipsoid, respectively. , , These are the three semi-major axes of the ellipsoid; For an ellipse with the length of equation (2) short half shaft With the three semi-major axes of the ellipsoid , , The equation for the projection relationship is: ， （4） in: ； 。 4. The visual estimation method for the three-dimensional dimensions of riprap according to claim 3, characterized in that: In step a.4), the stone samples are graded. The calculation method is as follows: ， in: M is the mass of each particle of the rockfill in the stone sample; This indicates the total number of riprap particles within this particle size range; This indicates the total number of riprap particles in (1); The expression for M is: ； in: It is the density of the rubble.

5. The visual estimation method for the three-dimensional dimensions of riprap according to claim 4, characterized in that: The specific implementation method of step 3) is as follows: The density of the rubble is Calculate the mass of each particle of the riprap. and gradation : （8） （9） in: This indicates the total number of riprap particles within this particle size range; This indicates the total number of riprap particles in (1).

Images