Particle contour feature algorithm based on tracer image and tracer simulation system

Abstract

The invention relates to the technical field of deep-sea mining simulation experiments, and discloses a particle contour feature algorithm based on a tracer image, which comprises the following steps of: after acquiring an image for simulating the movement of ore particles by using a fixed-view-angle camera, processing the image to extract a light spot parameter of a light spot projection in the image; after a contour feature calculation model is established, light spot parameters are input, and contour feature information is obtained through calculation; establishing an error relation model between the observation image and the model prediction image, importing the diameter of the simulated ore particles to obtain an error, and updating and optimizing the error; the invention also provides a tracing simulation system. An image contour processing mode is adopted, the edge of the simulated ore particle in the image is detected and processed, light spot parameters are extracted, a contour feature calculation model is established to process the light spot parameters, and the radius, the position, the posture and the edge point coordinates of the simulated ore particle are obtained based on the image; therefore, the contour feature of each simulated ore particle is accurately obtained, and accurate data is provided for motion correction analysis.

Description

Technical Field

[0001] The invention relates to the technical field of deep-sea mining simulation experiments, and in particular to a particle profile feature algorithm based on tracer images and a tracer simulation system. Background Art

[0002] The deep-sea mining simulated ore particle tracer experiment is to study the movement and behavior of ore particles in the deep-sea environment, especially the movement trajectory and mechanical properties under sediment erosion and turbid water conditions. This experiment is of great significance for optimizing mining equipment design, improving mining efficiency and environmental protection.

[0003] At present, deep-sea mining simulated ore particle tracing experiments usually use particle flow detection methods, such as velocimeters (such as particle image velocimetry, particle tracking velocimetry, etc.), fluorescence tracing, and imaging methods (magnetic resonance flow imaging, ultrasonic Doppler velocimetry, etc.). Although these methods can detect the movement of simulated ore particle flows, it is difficult to obtain the contour characteristics of a single simulated ore particle, and thus it is difficult to measure the motion posture (such as spin state) of the simulated ore particles, resulting in difficulty in accurately analyzing the motion information of the simulated ore particles, that is, it is difficult to accurately correct the force conditions of the simulated ore particles, which reduces the accuracy of the experiment. Summary of the invention

[0004] The purpose of the present invention is to provide a particle contour feature algorithm and a tracer simulation system based on tracer images, so as to solve the technical problem in the prior art that it is difficult to obtain the contour features of simulated ore particles and it is difficult to accurately obtain the motion posture of simulated ore particles, resulting in difficulty in accurately analyzing the motion information of simulated ore particles and difficulty in accurately correcting the stress conditions of simulated ore particles.

[0005] In order to solve the above technical problems, the present invention specifically provides the following technical solutions: A particle contour feature algorithm based on tracer images includes the following steps: Step 100: After collecting an image simulating the movement of ore particles using a fixed-viewing angle camera, grayscale processing is performed on the image and light point detection and light spot fitting are performed in sequence to extract light spot parameters of the light spot projection in the image; Step 200, establishing a contour feature calculation model for simulated ore particles, importing the spot parameters into the contour feature calculation model, so as to calculate and obtain the radius and position of the simulated ore particles in the image, the posture of the simulated ore particles, and the coordinates of the edge points of the simulated ore particles, thereby obtaining contour feature information of the simulated ore particles; Step 300: Establish an error relationship model between the observed image and the model predicted image, import the diameter of the simulated ore particles into the error relationship model to estimate the error, and iteratively update the model parameters using the partial derivatives of the error, and use the weighted least squares method to reduce the impact of outliers.

[0006] As a preferred solution of the present invention, in step 100, the specific method of using light spot detection and light spot fitting to extract the light spot parameters of the light spot projection in the image is: A partial differential equation of anisotropic diffusion is established to perform smoothing and edge detection on the image. The partial differential equation of anisotropic diffusion is: ; in, is the image at time The gray value of and represents the spatial coordinates of the image, represents the diffusion time, is the diffusion coefficient; is the gray value gradient of the light spot, is the divergence; The light spot detection is performed using a second-order partial differential operator, which is: ; Find a set of points that satisfy the spot equation such that: ; in, is the eigenvalue.

[0007] As a preferred solution of the present invention, in step 200, the specific method of establishing the contour feature calculation model of the simulated ore particles is: Establish the light spot parameter equation under the perspective projection model; Establish equations for calculating the radius and position of simulated ore particles; Establishing the posture equations for calculating and simulating ore particles; Establish edge coordinate equations for calculating simulated ore particles; The relationship equations, radius and position equations, posture equations and edge coordinate equations are combined to form a contour feature calculation model.

[0008] As a preferred solution of the present invention, the specific method of establishing the light spot parameter equation under the perspective projection model is: Define the gray value gradient field of the light spot: ; According to the gradient field information of the light spot, the parameterized equation of the light spot is established: ; And establish the geometric parameterized equation between the light spot and the camera: ; Among them, for the The geometric center coordinates are , is the major axis length, is the circular spot diameter, is the minor axis length, φ is the spot rotation angle, is the focal length of the camera, is the distance from the light spot to the camera, is the angle between the spot normal vector and the camera optical axis; in, and Obtained by image grayscale value, and , according to the conversion formula from spherical coordinate system to rectangular coordinate system: ; Since the normal vector is a unit vector, , then the normal vector of each point on the spot is It is expressed as: ; Introducing the camera intrinsic matrix : ; in, and The camera is and The focal length of the direction, and Camera principal point coordinates; The plane coordinates are normalized to get: ; Among them, the normalized The plane coordinates of the light spot are ; For the spot space point , and its projection on the image plane The following relations are satisfied: ; Therefore, the spatial coordinates of the light spot are: ; Among them, The spatial coordinates of the light spot .

[0009] As a preferred embodiment of the present invention, the specific method of establishing the radius and position equation for calculating the simulated ore particles is: Establish the spherical equation of each spot on the simulated ore particle: ; Among them, the spatial coordinates of the simulated ore particles , the radius of the simulated ore particle is .

[0010] Establish the normal vector constraint equation: According to the relationship between the normal vector and the center of the sphere, we can get: ; After finishing, we get: ; in, is an unknown scaling factor; Substituting the above formula into the spherical equation, we get:

[0011] Therefore, the parameter equation of the light spot and simulated ore particles is: ; Solve for the radius and position of the simulated ore particles: Combining all the light spots with the simulated ore particle parameter equation, we can get: ; Will and As variables, an overdetermined set of equations is constructed, and the least squares method is used to solve the set of equations to obtain the spatial coordinates of the simulated ore particles. and radius .

[0012] As a preferred solution of the present invention, the specific method of establishing the posture equation for calculating the simulated ore particles is: Using the normals of each point on the sphere, calculate the posture of the sphere, and differentiate the surface normal to obtain the surface normal vector: ; The posture of simulated ore particles can be determined by the angle between the normal and the line of sight. express: ; in, is the sight direction; So the angle between the normal and the line of sight is It can be expressed as: .

[0013] As a preferred solution of the present invention, the specific method of establishing the edge coordinate equation for calculating the simulated ore particles is: The edge of the image is detected using a multi-level edge detection algorithm: ; Edges occur at gradient maxima: ; By solving the partial differential form of the spherical equation and the camera model equation, the image coordinates of the edge points are obtained : ; The spatial coordinates of the edge points can be solved using the spherical equation .

[0014] As a preferred solution of the present invention, in step 300, the specific method of establishing the error relationship model between the observed image and the model predicted image is: Establish the error function equation: ; in, is the gray value of the observed image, is the grayscale value of the image predicted by the model; The partial derivative form of the error function equation with respect to the simulated ore particle radius R is: .

[0015] In order to solve the above technical problems, the present invention further provides the following technical solutions: A tracer simulation system applied to the above-mentioned particle contour feature algorithm based on tracer images, comprising: A water tank, wherein a thermostat is provided inside the water tank for maintaining a constant water temperature inside the water tank to simulate a deep sea environment; A high-speed short-wave infrared camera is arranged outside the water tank, and a detection end of the high-speed short-wave infrared camera is arranged perpendicular to the transparent side plate of the water tank to collect images inside the water tank; A porous infrared source ball is placed in the water tank to simulate ore particles. The porous infrared source ball has a built-in infrared source and releases infrared rays to its outer surface in a radiation manner to form an infrared light spot on its outer surface. Wherein, when the porous infrared source balls move disorderly, the high-speed short-wave infrared camera captures the moving pictures of the porous infrared source balls to output images for simulating the movement information of ore particles.

[0016] As a preferred solution of the present invention, the porous infrared source ball comprises an inner shell and an outer shell, a plurality of short-wave infrared generating tubes are arranged between the inner shell and the outer shell to be connected and fixed, and the plurality of short-wave infrared generating tubes are radiated from the outer shell with the center of the inner shell as the center. A plurality of holes are formed on the surface of the ball shell, the centrifugal end of the short-wave infrared generating tube is fixed in the hole, and its centrifugal end surface is exposed to the outside of the ball shell, and the ball shell is made of a heat-insulating material; A power supply battery pack is pre-installed in the inner shell of the sphere, and the power supply battery pack has a high specific heat capacity. The infrared rays of the power supply battery pack are transmitted to the centrifugal end surface of the short-wave infrared generating tube through the inner shell of the sphere to form an infrared spot on the outer surface of the outer shell of the sphere.

[0017] Compared with the prior art, the present invention has the following beneficial effects: The present invention adopts the method of image contour processing to detect and process the edges of simulated ore particles in the image and extract spot parameters, and establishes a contour feature calculation model to process the spot parameters, so as to obtain the radius and position, posture and edge point coordinates of the simulated ore particles based on the image, thereby accurately obtaining the contour features of each simulated ore particle and providing accurate data for motion correction analysis.

[0018] The present invention simulates ore particles by adopting a radiation infrared source ball, and sets a plurality of short-wave infrared generating tubes between the inner shell and the outer shell of the ball, so that the short-wave infrared generating tubes can conduct the infrared rays of the infrared source in the inner shell of the ball to the surface of the outer shell of the ball and release them in the form of a plurality of circular light spots. The high-speed short-wave infrared camera is used to collect image information without being disturbed by turbid water bodies, and effective control point information can be provided, thereby providing an accurate basis for edge computing and spin state calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the implementation methods of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for the implementation methods or the description of the prior art. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other implementation drawings can be derived from the provided drawings without creative work.

[0020] Figure 1 A flow chart of a particle contour feature algorithm based on tracer images provided in an embodiment of the present invention; Figure 2 A schematic diagram of the composition of a tracer simulation system provided by an embodiment of the present invention; Figure 3 A schematic diagram of the structure of a porous infrared source ball portion of a tracing simulation system provided by an embodiment of the present invention; Figure 4 A schematic diagram of the posture calculation process of the particle contour feature algorithm based on the tracer image provided in an embodiment of the present invention.

[0021] The numbers in the figure represent the following: 1-Porous infrared source ball; 11-ball inner shell; 12-ball outer shell; 13-short-wave infrared generating tube; 121-Hole. DETAILED DESCRIPTION

[0022] 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.

[0023] like Figure 1 , Figure 4 As shown, the present invention provides a particle contour feature algorithm based on tracer images, comprising the following steps: Step 100: After using a fixed-view camera to capture an image simulating the movement of ore particles, grayscale processing is performed on the image and light spot detection and spot fitting are performed in sequence to extract the spot parameters of the light spot projection in the image; Step 200, establishing a contour feature calculation model for simulated ore particles, importing the spot parameters into the contour feature calculation model, so as to calculate the radius and position of the simulated ore particles in the acquired image, the posture of the simulated ore particles, and the coordinates of the edge points of the simulated ore particles, thereby obtaining contour feature information of the simulated ore particles; Step 300: Establish an error relationship model between the observed image and the model predicted image, import the diameter of the simulated ore particles into the error relationship model to estimate the error, and iteratively update the model parameters using the partial derivatives of the error, and use the weighted least squares method to reduce the impact of outliers.

[0024] The contour feature algorithm of the present invention mainly obtains a grayscale image of the simulated ore particle motion image after a fixed-angle camera captures the image of the simulated ore particle motion, and performs photoelectric detection and spot fitting on it based on the grayscale image, so as to extract the spot parameters of the spot projection in the image, and then after establishing a contour feature calculation model of the simulated ore particles, the spot parameters are imported into the contour feature calculation model to obtain the radius and position, posture and edge point coordinates of the simulated ore particles, thereby obtaining the contour feature information of the simulated ore particles.

[0025] By establishing an error relationship model between the observed image and the model predicted image, the diameter of the simulated ore particles is introduced into the error relationship model to obtain the error. By iteratively updating the model parameters using the partial derivatives of the error and using the weighted least squares method, the influence of outliers is reduced to improve the accuracy of the contour feature algorithm.

[0026] Compared with the existing method of using particle flow detection, the contour feature algorithm of the present invention performs grayscale processing on the image after collecting the image of the simulated ore particles, and then performs light spot detection and spot fitting, so as to extract the spot parameters of the light spot projection in the image, and establish a contour feature calculation model. The spot parameters are imported into the contour feature calculation model to obtain the radius and position, posture and edge point coordinates of the simulated ore particles, thereby obtaining the contour features of the simulated ore particles.

[0027] In order to ensure the accuracy of the contour feature calculation model, an error relationship model between the observed image and the model prediction image is established, so that the diameter of the simulated ore particles is introduced into the error relationship model to obtain the error, and the model parameters are iteratively updated using the partial derivatives of the error. The weighted least squares method is used to reduce the influence of outliers and improve the accuracy of the contour feature algorithm, so as to obtain accurate contour features of the simulated ore particles, thereby improving the accuracy of the motion information analysis, and being able to more accurately correct the stress conditions of the simulated ore particles based on this contour feature.

[0028] In step 100, the specific method of using light spot detection and light spot fitting to extract the light spot parameters of the light spot projection in the image is: A partial differential equation of anisotropic diffusion is established to perform smoothing and edge detection on the image. The partial differential equation of anisotropic diffusion is: ; in, is the image at time The gray value of and represents the spatial coordinates of the image, Indicates the diffusion time. As time goes by, the image will gradually change, thus achieving smoothing and edge preservation. is the diffusion coefficient, which controls the rate of diffusion; is the gray value gradient of the light spot, is the divergence; The light spot detection is performed using a second-order partial differential operator, which is: ; Find a set of points that satisfy the spot equation such that: ; in, is the eigenvalue. By solving the above partial differential equation, the spot parameters of the light spot projection in the image can be extracted.

[0029] In this step, the diffusion coefficient is defined as: ; in, It is the threshold parameter that controls the edge and controls the sensitivity of the edge.

[0030] Specifically, the diffusion coefficient is the diffusion rate of pixel values ​​in the image, and the gray value Indicates the brightness value of each pixel in the image. In image smoothing (or denoising), the diffusion coefficient is used to control the diffusion rate of the pixel value. The anisotropic diffusion method can retain edge information and avoid blurred edges, thereby making the contour feature acquisition more accurate.

[0031] Of course, after the image has been processed to retain edge information, the position, size, posture and edge information of the simulated ore particles in the image need to be processed. The details are as follows: In step 200, the specific method of establishing the contour feature calculation model of the simulated ore particles is: Establish the light spot parameter equation under the perspective projection model; Establish equations for calculating the radius and position of simulated ore particles; Establishing the posture equations for calculating and simulating ore particles; Establish edge coordinate equations for calculating simulated ore particles; The relationship equations, radius and position equations, posture equations and edge coordinate equations are combined to form a contour feature calculation model.

[0032] The specific method of establishing the light spot parameter equation under the perspective projection model is: Define the gray value gradient field of the light spot: ; According to the gradient field information of the light spot, the parameterized equation of the light spot is established: ; And establish the geometric parameterization equation between the light spot and the camera. Specifically, when a circular hole is observed at an oblique angle, its projection on the image plane is an ellipse. The major axis and minor axis of the ellipse have the following relationship with the viewing angle: ; Among them, for the The geometric center coordinates are ; in, is the major axis length, is the circular spot diameter, is the minor axis length, is the light spot rotation angle; in, is the focal length of the camera, is the distance from the light spot to the camera, is the angle between the spot normal vector and the camera optical axis; Establish the geometric parameterization equation between the light spot and the camera: The long axis length of the spot is affected by perspective projection, and its size is related to the inclination angle of the camera optical axis. No relation, therefore When , the diameter of the projection circle of the light spot in the image can be obtained as the major axis length of the light spot: ; Due to the projection process, the length of the direction perpendicular to the rotation axis is shortened times, the minor axis length of the light spot is: ; Therefore, the geometric parameterization equation between the light spot and the camera can be expressed as: ; because and It can be obtained from the gray value of the image, so the spot inclination for: ; in, and Obtained through the image grayscale value, according to the conversion formula from spherical coordinate system to rectangular coordinate system: ; Since the normal vector is a unit vector, , then the normal vector of each point on the spot is It can be expressed as: ; Establish the spatial position equation of the light spot: In order to recover the spatial position of the light spot from the two-dimensional image and finally determine the projection position of the light spot in the image, the camera intrinsic parameter matrix is ​​introduced: ; in, and The camera is and The focal length of the direction, and Camera principal point coordinates; The plane coordinates are normalized to get: ; Among them, the normalized The plane coordinates of the light spot are ; For the spot space point , and its projection on the image plane The following relations are satisfied: ; Therefore, the spatial coordinates of the light spot are: ; Among them, The spatial coordinates of the light spot .

[0033] Specifically, and is the semi-major axis and semi-minor axis of the spot, and the spot diameter It refers to the actual diameter of the light spot projection in the image. This diameter can be calculated by the light spot parameters. The light spot may rotate in the image, so the rotation angle needs to be considered. , and the spot may be projected from the three-dimensional space to the two-dimensional image plane through perspective projection. Based on this, the relationship between the spot diameter and the spot axis is established, and the spot diameter can be calculated .

[0034] The specific method for establishing the radius and position equation for calculating the simulated ore particles is as follows: Establish the spherical equation of each spot on the simulated ore particle: ; Among them, the spatial coordinates of the simulated ore particles , the radius of the simulated ore particle is .

[0035] Establish the normal vector constraint equation: According to the relationship between the normal vector and the center of the sphere, we can get: ; After finishing, we get: ; in, is an unknown scaling factor; Substituting the above formula into the spherical equation, we get:

[0036] Therefore, the parameter equation of the light spot and simulated ore particles is: ; Solve for the radius and position of the simulated ore particles: Combining all the light spots with the simulated ore particle parameter equation, we can get: ; Will and As variables, an overdetermined set of equations is constructed, and the least squares method is used to solve the set of equations to obtain the spatial coordinates of the simulated ore particles. and radius .

[0037] The specific method of establishing the posture equation for calculating the simulated ore particles is as follows: Using the normals of each point on the sphere, calculate the posture of the sphere, and differentiate the surface normal to obtain the surface normal vector: ; The posture of simulated ore particles can be determined by the angle between the normal and the line of sight. express: ; in, is the sight direction; So the angle between the normal and the line of sight is It can be expressed as: .

[0038] Specifically, the surface normal is one of the key information for determining the posture of an object. The surface normal can be calculated through the spot parameters extracted by spot fitting, wherein the normal of the spot can be obtained by gradient calculation.

[0039] The angle between the normal direction and the line of sight can be used to determine the attitude parameters of an object. Commonly used representation methods include rotation matrix, Euler angle and quaternion.

[0040] The specific method for establishing the edge coordinate equation for calculating the simulated ore particles is as follows: The edge of the image is detected using a multi-level edge detection algorithm: ; The edge appears at the maximum gradient, that is: ; By solving the partial differential form of the spherical equation and the camera model equation, the image coordinates of the edge points are obtained : ; The spatial coordinates of the edge points can be solved using the spherical equation , then project the spatial coordinates back to the image and verify the relationship between the edge point and the sphere, that is, verify whether the edge point falls on the sphere.

[0041] Specifically, after edge detection, the edge appears at the maximum gradient value, so as to determine the edge information of the simulated ore particles in the image, and calculate the edge points of the simulated ore particles to obtain the image coordinates of the edge points, thereby obtaining the edge contour information of the simulated ore particles.

[0042] In step 300, the specific method of establishing the error relationship model between the observed image and the model predicted image is: Establish the error function equation: ; in, is the gray value of the observed image, is the grayscale value of the image predicted by the model; The partial derivative form of the error function equation with respect to the simulated ore particle radius R is: ; By calculating the partial derivatives of the error, we can use optimization algorithms to adjust the model parameters to minimize the error and improve the accuracy of the model.

[0043] Specifically, after establishing the error relationship model between the observed image and the model predicted image, the actual radius of the simulated ore particles can be input for error estimation, and the model iteration processing and outlier processing can be performed based on the error estimation to improve the accuracy of the model.

[0044] like Figure 2 , Figure 3 As shown, based on the particle contour feature algorithm based on the tracer image, a tracer simulation system is provided, including: A water tank is provided with a thermostat to maintain a constant water temperature inside the water tank to simulate a deep-sea environment; A high-speed short-wave infrared camera is arranged outside the water tank, and a detection end of the high-speed short-wave infrared camera is arranged perpendicular to a transparent side plate of the water tank to collect images inside the water tank; A porous infrared source ball 1 is placed in the water tank to simulate ore particles. The porous infrared source ball 1 has a built-in infrared source and releases infrared rays to its outer surface in a radiation manner to form an infrared spot on its outer surface. When the porous infrared source ball 1 moves disorderly, the high-speed short-wave infrared camera captures the moving picture of the porous infrared source ball 1 to output an image for simulating the movement information of ore particles.

[0045] The tracer simulation system of the present invention mainly utilizes a porous infrared source ball 1 to release infrared rays in the liquid of a water tank, which is then captured by a high-speed short-wave infrared camera outside the water tank. Due to the use of infrared acquisition, it is not disturbed by turbid water. During the capturing process, the porous infrared source ball 1 forms an infrared spot on its surface by radiation, so that the captured image contains an infrared spot pattern of the porous infrared source ball 1. Based on the infrared spot patterns of two porous infrared source balls 1 within the interval time, the posture and spin state of the porous infrared source ball 1 can be analyzed and obtained to provide more accurate data for motion correction.

[0046] Among them, the constant temperature device is to prevent multiple simulated ore particles from affecting the water temperature, so as to maintain the signal difference between the ambient temperature and the observed object and improve the observation effect.

[0047] Specifically, Figure 3 As shown, the porous infrared source ball 1 includes an inner shell 11 and an outer shell 12, and a plurality of short-wave infrared generating tubes 13 are arranged between the inner shell 11 and the outer shell 12 to be connected and fixed, and the plurality of short-wave infrared generating tubes 13 are radiated and distributed from the center of the inner shell 11 and the outer shell 12; A plurality of holes 121 are formed on the surface of the ball shell 12, and the centrifugal end of the short-wave infrared generating tube 13 is fixed in the hole 121, and its centrifugal end surface is exposed to the outside of the ball shell 12, and the ball shell 12 is made of a heat-insulating material; A power supply battery pack is pre-installed in the inner shell 11 of the ball, and the power supply battery pack is connected to the short-wave infrared generating tube. The power supply battery pack passes through the inner shell 11 of the ball to supply power to the short-wave infrared generating tube 13 to form an infrared spot on the outer surface of the outer shell 12 of the ball.

[0048] Through this porous infrared source ball 1, the insulated ball shell 12 prevents the observation of the infrared source from being disturbed and maintains the observation effect. Also, due to the simple structure, the power supply battery pack can be quickly replaced when the power is insufficient.

[0049] The cooperation between the end face of the short-wave infrared generating tube 13 at the hole 121 and the high-speed short-wave infrared camera can monitor the posture of the simulated ore particles through the position change of the hole 121, and can also obtain the position of the simulated ore particles through the position change of the hole 121 and the aforementioned contour feature algorithm, thereby obtaining the trajectory change of the simulated ore particles and obtaining the kinematic and dynamic information of the simulated ore particles.

[0050] In this process, the circular infrared spots formed by multiple short-wave infrared generating tubes 13 at the holes 121 can be used as control points in the image. After obtaining the contour features, the porous infrared source sphere 1 in its two time states is aligned based on the edge information of the contour features. The spin state of the porous infrared source sphere 1 can be accurately calculated based on the offset of the two control points, thereby providing more accurate data for motion analysis and motion correction.

[0051] The above embodiments are only exemplary embodiments of the present application and are not intended to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and protection scope of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the protection scope of the present application.

Claims

1. A particle contour feature algorithm based on tracer images, characterized in that: The steps include: Step 100: After collecting an image simulating the movement of ore particles using a fixed-viewing angle camera, grayscale processing is performed on the image and light point detection and light spot fitting are performed in sequence to extract light spot parameters of the light spot projection in the image; Step 200, establishing a contour feature calculation model for simulated ore particles, importing the spot parameters into the contour feature calculation model, so as to calculate and obtain the radius and position of the simulated ore particles in the image, the posture of the simulated ore particles, and the coordinates of the edge points of the simulated ore particles, thereby obtaining contour feature information of the simulated ore particles; Step 300: Establish an error relationship model between the observed image and the model predicted image, import the diameter of the simulated ore particles into the error relationship model to estimate the error, and iteratively update the model parameters using the partial derivatives of the error, and use the weighted least squares method to reduce the impact of outliers.

2. The particle contour feature algorithm based on tracer image according to claim 1, characterized in that: In step 100, the specific method of using light spot detection and light spot fitting to extract the light spot parameters of the light spot projection in the image is: A partial differential equation of anisotropic diffusion is established to perform smoothing and edge detection on the image. The partial differential equation of anisotropic diffusion is: ； in, is the image at time The gray value of and represents the spatial coordinates of the image, represents the diffusion time, is the diffusion coefficient; is the gray value gradient of the light spot, is the divergence; The light spot detection is performed using a second-order partial differential operator, which is: ； Find a set of points that satisfy the spot equation such that: ； in, is the eigenvalue.

3. The particle contour feature algorithm based on tracer image according to claim 2, characterized in that: In step 200, the specific method of establishing the contour feature calculation model of the simulated ore particles is: Establish the light spot parameter equation under the perspective projection model; Establish equations for calculating the radius and position of simulated ore particles; Establishing the posture equations for calculating and simulating ore particles; Establish edge coordinate equations for calculating simulated ore particles; The relationship equations, radius and position equations, posture equations and edge coordinate equations are combined to form a contour feature calculation model.

4. The particle contour feature algorithm based on tracer images according to claim 3 is characterized in that: The specific method of establishing the light spot parameter equation under the perspective projection model is: Define the gray value gradient field of the light spot: ； According to the gradient field information of the light spot, the parameterized equation of the light spot is established: ； And establish the geometric parameterized equation between the light spot and the camera: ； Among them, for the The geometric center coordinates are , is the major axis length, is the circular spot diameter, is the minor axis length, is the light spot rotation angle, is the focal length of the camera, is the distance from the light spot to the camera, is the angle between the light spot normal vector and the camera optical axis; in, and Obtained by image grayscale value, and , according to the conversion formula from spherical coordinate system to rectangular coordinate system: ； Since the normal vector is a unit vector, , then the normal vector of each point on the spot is It is expressed as: ； Introducing the camera intrinsic matrix : ； in, and The camera is and The focal length of the direction, and Camera principal point coordinates; The plane coordinates are normalized to get: ； Among them, the normalized The plane coordinates of the light spot are ; For the spot space point , and its projection on the image plane The following relations are satisfied: ； Therefore, the spatial coordinates of the light spot are: ； Among them, The spatial coordinates of the light spot .

5. The particle contour feature algorithm based on tracer images according to claim 4 is characterized in that: The specific method of establishing the equation for calculating the radius and position of simulated ore particles is: Establish the spherical equation of each spot on the simulated ore particle: ； Among them, the spatial coordinates of the simulated ore particles , the radius of the simulated ore particle is ; Establish the normal vector constraint equation: According to the relationship between the normal vector and the center of the sphere, we can get: ; Arrange to get: ;in, is an unknown proportional factor; substituting the above formula into the spherical equation yields: Therefore, the parameter equation of the light spot and simulated ore particles is: ; Solve the radius and position of the simulated ore particles: Combining all the light spots with the simulated ore particle parameter equations, we can get: ;Will and As variables, an overdetermined set of equations is constructed, and the least squares method is used to solve the set of equations to obtain the spatial coordinates of the simulated ore particles. and radius .

6. The particle contour feature algorithm based on tracer images according to claim 4, characterized in that: The specific method of establishing the posture equation for calculating the simulated ore particles is: Using the normals of each point on the sphere, calculate the posture of the sphere, and differentiate the surface normal to obtain the surface normal vector: ； The posture of simulated ore particles can be determined by the angle between the normal and the line of sight. express: ； in, is the sight direction; So the angle between the normal and the line of sight is It can be expressed as: 。 7. The particle contour feature algorithm based on tracer images according to claim 4, characterized in that: The specific method for establishing the edge coordinate equation for calculating the simulated ore particles is: The edge of the image is detected using a multi-level edge detection algorithm: ； Edges occur at gradient maxima: ； By solving the partial differential form of the spherical equation and the camera model equation, the image coordinates of the edge points are obtained : ； The spatial coordinates of the edge points can be solved using the spherical equation .

8. The particle contour feature algorithm based on tracer images according to claim 5, characterized in that: In step 300, the specific method of establishing the error relationship model between the observed image and the model predicted image is: Establish the error function equation: ； in, is the gray value of the observed image, is the grayscale value of the image predicted by the model; The partial derivative form of the error function equation with respect to the simulated ore particle radius R is: 。 9. A tracer simulation system applied to the particle contour feature algorithm based on tracer images as claimed in any one of claims 1 to 8, characterized in that: include: A water tank, wherein a thermostat is provided inside the water tank for maintaining a constant water temperature inside the water tank to simulate a deep sea environment; A high-speed short-wave infrared camera is arranged outside the water tank, and a detection end of the high-speed short-wave infrared camera is arranged perpendicular to the transparent side plate of the water tank to collect images inside the water tank; A porous infrared source ball (1) is placed in the water tank to simulate ore particles, wherein the porous infrared source ball (1) has a built-in infrared source and releases infrared rays to its outer surface in a radiation manner to form an infrared light spot on its outer surface; When the porous infrared source ball (1) moves disorderly, the high-speed short-wave infrared camera captures the moving picture of the porous infrared source ball (1) to output an image for simulating the movement information of ore particles.

10. A tracer simulation system according to claim 9, characterized in that: The porous infrared source sphere (1) comprises an inner shell (11) and an outer shell (12), a plurality of short-wave infrared generating tubes (13) are arranged between the inner shell (11) and the outer shell (12) and connected and fixed to each other, and the plurality of short-wave infrared generating tubes (13) are radiated and distributed from the outer shell (12) with the center of the inner shell (11) being the center of the inner shell (11); A plurality of holes (121) are formed on the surface of the ball shell (12), the centrifugal end of the short-wave infrared generating tube (13) is fixed in the hole (121), and its centrifugal end surface is exposed to the outside of the ball shell (12), and the ball shell (12) is made of a heat-insulating material; A power supply battery pack is pre-installed in the inner shell (11) of the sphere, the power supply battery pack is connected to a short-wave infrared generating tube, and the power supply battery pack passes through the inner shell (11) of the sphere to supply power to the short-wave infrared generating tube (13), so as to form an infrared light spot on the outer surface of the outer shell (12).

Images