A visually assisted method for fine-tuning zircon particle size

Through visually assisted detection methods, the detection accuracy problem caused by adhesion and stacking of zircon particles is solved. Through edge detection and local occlusion analysis, the accuracy of zircon particle size fine detection is improved, ensuring product quality and Productivity.

CN118429372BActive Publication Date: 2024-11-05SHANDONG LONGBAI TITANIUM IND TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410473516.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-04-19
Publication Date
2024-11-05
Estimated Expiration
2044-04-19

AI Technical Summary

Technical Problem

The number of zircon particles is large and easy to stick and stack, resulting in low accuracy in particle size refinement detection, which seriously affects product quality and production efficiency.

Method used

Using visually assisted detection method, the edge lines and stacked boundary lines of zircon particles are obtained through edge detection, the boundary line membership and local occlusion degree are calculated, and the occluded pixel points are predicted, and the particle size characteristic confidence and particle size refinement index are obtained. , improve detection accuracy.

Benefits of technology

It improves the estimation efficiency and accuracy of the degree of occlusion of zircon particles, enhances the accuracy of particle size refinement detection, and ensures the reliability of zircon particle size detection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118429372B_ABST
    Figure CN118429372B_ABST
Patent Text Reader

Abstract

The present invention relates to the field of image processing technology, and proposes a visually assisted zircon grain size refinement detection method, including: obtaining the edge line of the zircon grain, and then obtaining the stacking boundary line, boundary line membership and specific pixel points; obtaining the predicted pixel point sequence according to the stacking boundary line; obtaining the occluded pixel points and the prediction accuracy according to the predicted pixel point sequence, and then obtaining the local occlusion degree; combining the boundary line membership and the number of pixels inside the zircon grain to obtain the particle size feature confidence, and then obtaining the effective zircon grain; obtaining the collection particle size of the effective zircon grain, and then obtaining the particle size refinement index; obtaining the particle size refinement grade according to the particle size refinement index. The present invention aims to solve the problem that the traditional particle size refinement detection method has low accuracy when detecting zircon.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a visually assisted zircon particle size refinement detection method. Background Art

[0002] Zircon is extremely resistant to high temperatures, with a melting point of 2750°C, and is resistant to acid corrosion. 80% of the world's zircon is directly used in the foundry industry, ceramics, glass industry, and the manufacture of refractory materials. Zircon has lower thermal expansion, higher thermal conductivity, and stronger chemical stability than other common refractory materials. Therefore, high-quality zircon has good bonding properties with various other adhesives and is used in the foundry industry. Among them, in the production of zircon, the degree of refinement of particle size is one of the key parameters, which directly affects the quality of the product and production efficiency.

[0003] However, there are a large number of zircon particles, which are prone to adhesion and stacking. During the fine-grained detection of zircon particle size, they are easily affected by the mutual obstruction of many zircon particles, resulting in inaccurate detection of the obscured zircon particle size characteristics, which seriously affects the accuracy of subsequent fine-grained detection of zircon particle size. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a visually assisted zircon particle size refinement detection method to solve the problem that the existing particle size refinement detection method is not accurate enough when detecting zircon. The technical solution adopted is as follows:

[0005] An embodiment of the present invention provides a visually assisted method for fine-tuning the particle size of zircon, the method comprising the following steps:

[0006] Acquire zircon sample images;

[0007] Perform edge detection on zircon sample images to obtain zircon particles and edge lines of zircon particles, and then obtain stacking boundary lines of zircon particles; obtain the boundary line membership of the stacking boundary line according to the positional relationship between the pixel points on the stacking boundary line and the edge line of the zircon particles; obtain the predicted pixel point sequences at the two end points of the stacking boundary line according to the pixel points on the edge line of the zircon particles and the two end points of the stacking boundary line; obtain the occluded pixel points and the prediction accuracy of the occluded pixel points according to the predicted pixel point sequences at the two end points; obtain the local occlusion degree according to the prediction accuracy of the occluded pixel points, the distance between the occluded pixel points and the belonging boundary line, and the distance between the pixel points on the stacking boundary line and the belonging boundary line; obtain the particle size feature confidence of the zircon particles according to the boundary line membership of the stacking boundary line, the local occlusion degree and the number of pixels inside the zircon particles, and then obtain effective zircon particles;

[0008] Obtain the collection particle size of effective zircon particles; obtain the particle size refinement index of the zircon sample according to the collection particle size of the effective zircon particles; obtain the particle size refinement grade of the zircon sample according to the particle size refinement index.

[0009] Furthermore, the edge detection of the zircon sample image is performed to obtain the zircon particles and the edge lines of the zircon particles, and then the stacking boundary lines of the zircon particles are obtained, including the specific method of:

[0010] The edge detection operator is used to detect the edge of the zircon sample image to obtain the edge line;

[0011] The inner area of ​​the closed edge line was recorded as zircon grain;

[0012] The closed edge line was recorded as the edge line of the zircon grain;

[0013] Each zircon grain is recorded as a zircon grain to be analyzed;

[0014] The overlapping parts of the edge line of the zircon particle to be analyzed and the edge lines of the surrounding zircon particles are respectively recorded as a stacking boundary line of the zircon particle to be analyzed.

[0015] Furthermore, the specific method of obtaining the borderline membership of the stacking borderline according to the positional relationship between the pixel points on the stacking borderline and the edge line of the zircon particle includes:

[0016] Each stacking boundary line of the zircon grains to be analyzed is recorded as a stacking boundary line to be analyzed;

[0017] Acquire two endpoints of the stack boundary line to be analyzed, and record the line between the two endpoints of the stack boundary line to be analyzed as the subordinate boundary line corresponding to the stack boundary line to be analyzed;

[0018] The pixel points on the belonging boundary line located inside the edge line of the zircon grain to be analyzed are recorded as belonging pixel points;

[0019] The number of all pixels on the belonging boundary line is recorded as the number of pixels on the belonging boundary;

[0020] The ratio of the number of affiliated pixels to the number of affiliated boundary pixels is recorded as the membership ratio of the stack boundary line to be analyzed;

[0021] Assigning the borderline membership of the stacked borderline whose membership ratio is greater than the preset ratio to 1;

[0022] The border line membership of the stacked border lines whose membership ratio is less than or equal to the preset ratio is assigned a value of 0.

[0023] Furthermore, the method of obtaining predicted pixel point sequences at the two endpoints of the stacking boundary line respectively according to the pixel points on the edge line of the zircon particle and the two endpoints of the stacking boundary line includes the following specific methods:

[0024] The pixel points on the edge line of the zircon particle to be analyzed except all stacking boundary lines are recorded as specific pixel points of the zircon particle to be analyzed;

[0025] The two end points of the to-be-analyzed stacking boundary line of the to-be-analyzed zircon grains are respectively recorded as to-be-analyzed end points;

[0026] Among the specific pixels of the zircon grain to be analyzed, a preset number of pixels closest to the end point to be analyzed are recorded as predicted pixels;

[0027] According to the coordinates of all predicted pixels, the time series prediction model is used to obtain the coordinates of the number of pixels belonging to the boundary pixels;

[0028] The pixel points corresponding to the coordinates of the number of pixel points belonging to the boundary pixel points are recorded as predicted pixel points;

[0029] Arrange the predicted pixel points in the order of prediction to obtain the predicted pixel point sequence at the endpoint to be analyzed.

[0030] Furthermore, the method of obtaining the blocked pixel and the prediction accuracy of the blocked pixel according to the predicted pixel sequence at the two endpoints includes the following specific methods:

[0031] The two endpoints of the stack boundary line to be analyzed are recorded as the forward endpoint and the reverse endpoint respectively;

[0032] The reverse sequence of the predicted pixel point sequence at the reverse endpoint is recorded as the reverse predicted pixel point sequence at the reverse endpoint;

[0033] The two pixels with the same sequence number in the predicted pixel sequence at the forward endpoint and the reverse predicted pixel sequence at the reverse endpoint are recorded as a pixel pair;

[0034] Each pixel pair is recorded as a pixel pair to be analyzed;

[0035] The coordinate center points of the two pixels in the pixel pair to be analyzed are recorded as the blocked pixels;

[0036] The Euclidean distance between two pixels in the pixel pair to be analyzed is recorded as the distance of the pixel pair to be analyzed;

[0037] The sum of the distances of all pixel pairs is recorded as the prediction accuracy of the occluded pixel.

[0038] Furthermore, the method of obtaining the local occlusion degree according to the prediction accuracy of the occluded pixel, the distance between the occluded pixel and the belonging boundary line, and the distance between the pixel on the stacking boundary line and the belonging boundary line includes the following specific methods:

[0039] The sum of the shortest distances from all blocked pixels to their belonging boundary lines is recorded as the predicted partial blocking area;

[0040] The sum of the shortest distances from all pixels on the stacking boundary line to the belonging boundary line is recorded as the fixed partial occlusion area;

[0041] The difference between the number 1 and the power with the natural constant as the base and the opposite number of the prediction accuracy as the exponent is recorded as the occlusion weight;

[0042] The product of the prediction accuracy and the fixed partial occlusion area is recorded as the fixed occlusion degree;

[0043] The product of the predicted partial occlusion area and the occlusion weight is recorded as the predicted occlusion degree;

[0044] The sum of the fixed occlusion degree and the predicted occlusion degree is recorded as the local occlusion degree of the stack boundary line to be analyzed.

[0045] Furthermore, the specific method of obtaining the particle size characteristic confidence of the zircon particles according to the boundary line membership of the stacked boundary line, the local occlusion degree and the number of pixels inside the zircon particles, and then obtaining the effective zircon particles includes:

[0046] The difference between the number 1 and the degree of membership of the boundary line is recorded as the occlusion judgment value of the stacked boundary line to be analyzed;

[0047] The product of the local occlusion degree of the stack boundary line to be analyzed and the occlusion judgment value is recorded as the overall occlusion degree;

[0048] The ratio of the sum of the overall occlusion degrees of all stacked boundary lines to the number of pixel points inside the zircon grain to be analyzed is recorded as the occlusion ratio of the zircon grain to be analyzed;

[0049] The minimum value of the shielding ratio and the number 1 is recorded as the shielding characteristic value of the zircon grain to be analyzed;

[0050] The difference between the number 1 and the obstruction characteristic value is taken as the particle size characteristic confidence of the zircon particles to be analyzed;

[0051] Zircon particles whose particle size characteristic confidence is not equal to 0 are recorded as valid zircon particles.

[0052] Furthermore, the method of obtaining the effective zircon particle size includes:

[0053] The effective zircon particles are tested using the oriented maximum intercept method to obtain the oriented maximum intercept of the effective zircon particles in each direction;

[0054] The maximum value of the maximum oriented intercept of the effective zircon particles in each direction is taken as the collected particle size of the effective zircon particles.

[0055] Furthermore, the specific method of obtaining the particle size refinement index of the zircon sample according to the collected particle size of the effective zircon particles includes:

[0056] The product of the mean of the collected particle sizes of all effective zircon particles and the standard deviation of the collected particle sizes of all effective zircon particles is recorded as the zircon particle size difference;

[0057] The power of the natural constant as the base and the inverse number of the zircon particle size difference as the exponent is recorded as the particle size refinement index of the zircon sample.

[0058] Furthermore, the specific method of obtaining the particle size refinement grade of the zircon sample according to the particle size refinement index includes:

[0059] Classifying the particle size refinement level of the zircon sample whose particle size refinement index is less than or equal to the first preset threshold as low;

[0060] Classifying the grain refinement level of the zircon sample whose grain refinement index is between the first preset threshold and the second preset threshold as medium;

[0061] The grain refinement level of the zircon sample having a grain refinement index greater than or equal to the second preset threshold is classified as high.

[0062] The beneficial effects of the present invention are as follows: the present invention obtains a stacking boundary line according to the mutual occlusion of zircon particles; constructs a boundary line membership according to the positional relationship between the pixel points on the line segment formed by connecting the end points of the stacking boundary line and the edge line of the zircon particle, determines whether the stacking boundary line is the real edge of the zircon particle, that is, whether the zircon particle is occluded by other zircon particles at the stacking boundary line, thereby improving the efficiency of estimating the occlusion degree of the zircon particle; predicts the occluded pixel points according to the pixel points at the two end points of the stacking boundary line, thereby improving the prediction accuracy of the occluded pixel points; and by analyzing the occluded pixel points of the zircon particles, the occluded pixel points are predicted. The degree of occlusion needs to be judged based on its own specific edge and the occluded edge. Different weights are given to the occluded pixels and the pixels on the specific edge line according to the prediction accuracy, and the local occlusion degree of the zircon particles is calculated, thereby improving the accuracy of estimating the local occlusion degree of the zircon particles. By analyzing the influence of the occlusion degree of zircon particles on the particle size calculation, the particle size characteristic confidence is constructed according to the local occlusion degree, and the effective zircon particles are obtained. Then, the particle size refinement degree of the zircon sample is determined according to the collected particle size of the effective zircon particles, thereby improving the accuracy of zircon particle size refinement detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0064] Figure 1 A schematic flow chart of a visually assisted zircon particle size refinement detection method provided by an embodiment of the present invention;

[0065] Figure 2 Schematic diagram of zircon particle stacking;

[0066] Figure 3 Schematic diagram of the blocked area of ​​zircon particles;

[0067] Figure 4 Schematic diagram of the measurement of the oriented maximum intercept of particles. DETAILED DESCRIPTION

[0068] 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.

[0069] See also Figure 1 , which shows a flow chart of a visually assisted zircon grain size refinement detection method provided by an embodiment of the present invention, the method comprising the following steps:

[0070] Step S001: collect an initial image of a zircon sample and preprocess it to obtain a zircon sample image.

[0071] Sampling is done from the zircon production line or finished zircon warehouse. First, the total number of zircon bags W is obtained, and N small bags of zircon samples are randomly selected for particle size refinement testing. The empirical value of N is Where floor() is a rounding function. Then, the CCD camera is used to collect the initial image of the zircon sample. Since the zircon particles are relatively compact and have similar colors, the edges may be fuzzy. Therefore, before fine-grained particle size detection of the zircon sample image, it is necessary to use bilateral filtering to enhance the edge of the initial image of the zircon sample to obtain the zircon sample image.

[0072] Step S002: According to the mutual occlusion of zircon particles, the stacking boundary line and the boundary line membership of the stacking boundary line are obtained, and then the predicted pixel point sequence is obtained; and the occluded pixel points and the prediction accuracy are obtained according to the predicted pixel point sequence.

[0073] Since the particle size characteristics of each zircon particle in the obtained zircon sample image are different, when detecting the refinement degree of zircon particle size, it is necessary to consider the particle size characteristics of each zircon particle as much as possible, and obtain an overall evaluation of the refinement degree of zircon particle size. The present invention first uses the canny edge detection operator to detect the edge of the zircon sample image to obtain the edge line, and the inner area of ​​each closed edge line is recorded as a zircon particle; the closed edge line is recorded as the edge line of the zircon particle, and the number x of zircon particles is obtained.

[0074] When zircon particles are blocked, they cannot be fully displayed in the image. In order to avoid inaccurate detection results caused by blocked zircon particles, it is necessary to give a smaller confidence level to the particle size characteristics of zircon particles with a higher degree of blockage.

[0075] When two zircon grains are stacked together, their edge pixels will overlap at the stacking boundary, e.g. Figure 2As shown in the figure, zircon particle C blocks zircon particle B and is blocked by zircon particle A. The dotted line indicates the blocked edge line. Compare the edge pixels of each zircon particle. When the edge pixels in a zircon particle overlap with the edge pixels of other zircon particles, the overlapping edge pixels are recorded as overlapping edge pixels. Since a zircon particle may be stacked with multiple zircon particles, the overlapping edge pixels may not be adjacent and need to be processed separately. First, connect the adjacent overlapping edge pixels into a stacking boundary line.

[0076] When a zircon particle is blocked by other zircon particles, its corresponding stacking boundary actually belongs to other zircon particles. When a zircon particle blocks other zircon particles, its corresponding stacking boundary actually belongs to the zircon particle. Therefore, it is possible to judge whether each stacking boundary belongs to the zircon particle based on the characteristics of the stacking boundary, and then measure the degree of obstruction.

[0077] Get the number of stacking boundary lines N contained in the i-th zircon grain (i) , i is an integer between 1 and x. The endpoints of the jth stacking boundary line of the i-th zircon grain are marked as A and B, respectively, and the value of j is in the range of 1 to N (i) The line segment formed by connecting the two end points is recorded as the belonging boundary line l AB , if the jth stacking boundary line belongs to the i-th zircon grain, such as Figure 2 The line segment a2 shown in the figure has points on the boundary line mostly located inside the edge line of the zircon grain; otherwise, Figure 2 As shown in the line segment a1, most of the points on the belonging boundary line are located outside the edge line of the zircon grain. AB The pixels located inside the edge line of the i-th zircon grain are recorded as belonging pixels, and a threshold t1 is set. When the ratio of the number of belonging pixels to the number of all pixels on the belonging boundary line exceeds the threshold t1, the j-th stacking boundary line is determined to be the true edge of the i-th zircon grain, and the boundary line membership degree bBL is used. (i,j) =1; otherwise, it is judged that the jth stacking boundary line is not the true edge of the i-th zircon grain, that is, the stacking boundary line formed by being blocked by other zircon grains, and the boundary line membership degree bBL is used. (i,j) =0 indicates that the threshold value t1 is empirically taken as 80%.

[0078] When a zircon particle is blocked by other zircon particles, although the blocked part cannot be directly observed by the human eye, the position of the blocked pixel can be predicted based on the position of the unblocked pixel, thereby determining the degree of blockage of the zircon particle.

[0079] If the position of the occluded pixel is predicted based on only one of the two endpoints, the position of the edge pixel of the occluded part may be inaccurate, and the last predicted edge pixel deviates from the other endpoint. Therefore, it is necessary to predict the position of the edge pixel of the occluded part based on the edge pixel points at both ends corresponding to the two endpoints.

[0080] Get the belonging boundary line l AB The number of all pixels n2 (i,j) , on the i-th zircon grain, the M edge pixels adjacent to the endpoints A and B are respectively used to form the predicted pixel sequence of the A end and the predicted pixel sequence of the B end in the order from far to near, and the empirical value of M is 10. Figure 2 As shown, the predicted pixel point sequences of end A and end B are F, E, D, C, A and J, I, H, G, B respectively.

[0081] According to the horizontal and vertical coordinates of the two predicted pixel point sequences, the ARIMA time series prediction model is used to predict the subsequent n2 (i,j) The horizontal and vertical coordinates of the k1th pixel point (the edge pixel point of the occluded part) are calculated, and the average of the two sequence prediction results is taken as the horizontal and vertical coordinates of the final predicted occluded pixel point. The horizontal and vertical coordinates of the k1th occluded pixel point predicted by the pixel point sequence predicted by end A are The horizontal and vertical coordinates of the k1th occluded pixel point predicted by the B-side predicted pixel sequence are The accuracy of the predicted position of the occluded pixel is (i,j) The prediction accuracy of the blocked pixel corresponding to the jth stacking boundary line of the i-th zircon grain can be expressed by the difference between the two sequence prediction results. (i,j) The calculation formula is:

[0082]

[0083] Among them, BAG (i,j) is the prediction accuracy of the blocked pixel corresponding to the jth stacking boundary line of the i-th zircon grain; are the horizontal and vertical coordinates of the k1th occluded pixel point predicted according to the pixel point sequence predicted by end A, They are respectively the (n2 (i,j) -k1) the horizontal and vertical coordinates of the occluded pixels.

[0084] Due to the different prediction directions, the k1th edge pixel of the occluded part predicted by the predicted pixel sequence of end A corresponds to the (n2th edge pixel of the occluded part predicted by the predicted pixel sequence of end B. (i,j)-k1) edge pixels. When the distance between corresponding pixels is larger, it means that the difference between the results obtained from the two sequences is larger, the prediction error is more likely, and the prediction accuracy is lower.

[0085] Step S003: Obtain the local occlusion degree according to the occluded pixels and the prediction accuracy, and then obtain the particle size feature confidence by combining the boundary line membership and the number of pixels inside the zircon particles, and obtain the valid zircon particles according to the particle size feature confidence.

[0086] like Figure 3 As shown, there are n1 (i,j) The distance between the k1th occluded pixel and the belonging boundary line is The distance from the k2th coincident edge pixel to the belonging boundary line is

[0087] Since the edge pixels (including the edge pixels of the blocked part and the overlapping part) are compact enough, the sum of the Euclidean distances from each edge pixel to the belonging boundary line can be approximated as the area of ​​the zircon grain blocked, reflecting the degree of blockage of the zircon grain at the stacking boundary line. Among them, the local blockage degree JCov of the i-th zircon grain at the j-th stacking boundary line is (i,j) The calculation formula is:

[0088]

[0089]

[0090] Among them, BAG (i,j) is the prediction accuracy of the blocked part corresponding to the jth stacking boundary line of the i-th zircon grain; w (i,j) is the occlusion weight; JCov (i,j) is the local shading degree of the jth stacking boundary line of the i-th zircon grain; e is a natural constant; From the k1th occluded pixel to the line segment l AB The shortest distance along the straight line; The k2th coincident edge pixel to line segment l AB The shortest distance along the straight line.

[0091] When the prediction accuracy of the occluded part is higher, the occlusion weight is higher. The greater the distance between each edge pixel and the line segment formed by the end point of the stacking boundary line, the higher the local occlusion degree of the zircon grain at the stacking boundary line. The occlusion degree is determined based on the actual edge and the predicted occluded edge. The respective occlusion weights are determined by the accuracy of the prediction, which improves the accuracy of estimating the local occlusion degree of the zircon grain.

[0092] The more each stacking boundary line on a zircon particle is blocked by other zircon particles, the higher the overall blocking degree of the zircon particle is. The result of using its particle size characteristics to detect the refinement degree of the zircon sample particle size is less reliable and the confidence is lower.

[0093] There are N in the edge of the i-th zircon grain (the unobstructed part) (i) The number of pixels inside the zircon particle can be used to approximate the area of ​​the unobstructed part of the zircon particle. Combined with the local obstruction degree at each stacking boundary line and its boundary line membership, the particle size characteristic confidence wb of the i-th zircon particle is obtained: (i) , the calculation formula is:

[0094]

[0095] Among them, min{} is used to obtain the minimum value in a set of data represented by the box; wb (i) is the particle size characteristic confidence bBL of the i-th zircon particle (i,j) is the boundary line membership of the jth stacking boundary line of the i-th zircon grain. When the value is 0, it means that the stacking boundary line is not the true edge of the i-th zircon grain, that is, the zircon grain at the stacking boundary line is blocked by other zircon grains; JCov (i,j) Nin is the local occlusion degree of the i-th zircon grain at the j-th stacking boundary line, representing the area of ​​the zircon grain that is occluded at the stacking boundary line; (i) is the number of pixels inside the i-th zircon grain; N (i) is the number of stacking boundary lines contained in the i-th zircon grain.

[0096] When the ratio of the area of ​​the blocked part to the area of ​​the unblocked part of a zircon particle is larger, the particle size characteristics of the zircon particle are less referenceable and the confidence of its particle size characteristics is lower. When the ratio of the area of ​​the blocked part to the area of ​​the unblocked part is greater than or equal to 1, the confidence of its particle size characteristics reaches the minimum value of 0. At this time, the zircon particle is determined to be an invalid zircon particle, and the particle size characteristics of the zircon particle are not considered when the particle size of the zircon sample is detected. The zircon particles whose particle size characteristic confidence is not 0 are valid zircon particles.

[0097] Step S004: Calculate the collected particle size of effective zircon particles, and then obtain the particle size refinement index of the zircon sample; and obtain the particle size refinement grade of the zircon sample according to the particle size refinement index.

[0098] The size of zircon particles is called particle size. Generally, the size of particles is expressed in diameter, so it is also called particle size. The directional maximum intercept refers to the maximum diameter of particles measured along a certain direction. It is suitable for the determination of the particle size of granular mineral particles or aggregates. The measurement diagram of the directional maximum intercept along the 0° direction is shown in the figure. Figure 4 Zircon particles are granular mineral particles. The oriented maximum intercept method can be used in the zircon sample image to detect the particle size characteristics of each zircon particle, and the maximum value of the oriented maximum intercept in each direction of the zircon particle is taken as the collected particle size of the zircon particle.

[0099] The mean value of the collected particle size of each effective zircon particle is V, and the standard deviation is σ. The particle size refinement index GJG of the measured zircon can be expressed as follows:

[0100]

[0101] Among them, GJG is the particle size refinement index of the measured zircon; is the mean of the particle sizes of all effective zircon particles; σ is the standard deviation of the particle sizes of all effective zircon particles.

[0102] When the mean and variance of the collected particle size of each zircon particle are smaller, it means that the zircon particles are small and evenly distributed, the overall particle size of zircon is more refined, and the particle size refinement index is larger.

[0103] According to the value range of the particle size refinement index, the thresholds t2 and t3 are set, and the particle size refinement of the zircon sample is divided into three levels: low, medium and high. That is, when GJG≤t2, the particle size refinement is low, and the particle size refinement level of the zircon sample is divided into low; when t2<GJG<t3, the particle size refinement level is medium, and the particle size refinement level of the zircon sample is divided into medium; when GJG≥t3, the particle size refinement level is high, and the particle size refinement level of the zircon sample is divided into high. The empirical values ​​of t2 and t3 are 0.3 and 0.6 respectively.

[0104] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A visually assisted zircon particle size refinement detection method, characterized in that: The method comprises the following steps: Acquire zircon sample images; Perform edge detection on zircon sample images to obtain zircon particles and edge lines of zircon particles, and then obtain stacking boundary lines of zircon particles; obtain the boundary line membership of the stacking boundary line according to the positional relationship between the pixel points on the stacking boundary line and the edge line of the zircon particles; obtain the predicted pixel point sequences at the two end points of the stacking boundary line according to the pixel points on the edge line of the zircon particles and the two end points of the stacking boundary line; obtain the occluded pixel points and the prediction accuracy of the occluded pixel points according to the predicted pixel point sequences at the two end points; obtain the local occlusion degree according to the prediction accuracy of the occluded pixel points, the distance between the occluded pixel points and the belonging boundary line, and the distance between the pixel points on the stacking boundary line and the belonging boundary line; obtain the particle size feature confidence of the zircon particles according to the boundary line membership of the stacking boundary line, the local occlusion degree and the number of pixels inside the zircon particles, and then obtain effective zircon particles; Obtain the collection particle size of effective zircon particles; obtain the particle size refinement index of the zircon sample according to the collection particle size of the effective zircon particles; obtain the particle size refinement grade of the zircon sample according to the particle size refinement index.

2. A visually assisted zircon particle size refinement detection method according to claim 1, characterized in that: The specific method of performing edge detection on the zircon sample image to obtain zircon particles and edge lines of the zircon particles, and then obtaining the stacking boundary line of the zircon particles, includes: The edge detection operator is used to detect the edge of the zircon sample image to obtain the edge line; The inner area of ​​the closed edge line was recorded as zircon grain; The closed edge line was recorded as the edge line of the zircon grain; Each zircon grain is recorded as a zircon grain to be analyzed; The overlapping parts of the edge line of the zircon particle to be analyzed and the edge lines of the surrounding zircon particles are respectively recorded as a stacking boundary line of the zircon particle to be analyzed.

3. A visually assisted zircon particle size refinement detection method according to claim 2, characterized in that: The specific method of obtaining the borderline membership of the stacking borderline according to the positional relationship between the pixel points on the stacking borderline and the edge line of the zircon grains includes: Each stacking boundary line of the zircon grains to be analyzed is recorded as a stacking boundary line to be analyzed; Acquire two endpoints of the stack boundary line to be analyzed, and record the line between the two endpoints of the stack boundary line to be analyzed as the subordinate boundary line corresponding to the stack boundary line to be analyzed; The pixel points on the belonging boundary line located inside the edge line of the zircon grain to be analyzed are recorded as belonging pixel points; The number of all pixels on the belonging boundary line is recorded as the number of pixels on the belonging boundary; The ratio of the number of affiliated pixels to the number of affiliated boundary pixels is recorded as the membership ratio of the stack boundary line to be analyzed; Assigning the borderline membership of the stacked borderline whose membership ratio is greater than the preset ratio to 1; The border line membership of the stacked border lines whose membership ratio is less than or equal to the preset ratio is assigned a value of 0.

4. A visually assisted zircon particle size refinement detection method according to claim 3, characterized in that: The specific method of obtaining predicted pixel point sequences at the two endpoints of the stacking boundary line respectively according to the pixel points on the edge line of the zircon grain and the two endpoints of the stacking boundary line includes: The pixel points on the edge line of the zircon particle to be analyzed except all stacking boundary lines are recorded as specific pixel points of the zircon particle to be analyzed; The two end points of the to-be-analyzed stacking boundary line of the to-be-analyzed zircon grains are respectively recorded as to-be-analyzed end points; Among the specific pixels of the zircon grain to be analyzed, a preset number of pixels closest to the end point to be analyzed are recorded as predicted pixels; According to the coordinates of all predicted pixels, the time series prediction model is used to obtain the coordinates of the number of pixels belonging to the boundary pixels; The pixel points corresponding to the coordinates of the number of pixel points belonging to the boundary pixel points are recorded as predicted pixel points; Arrange the predicted pixel points in the order of prediction to obtain the predicted pixel point sequence at the endpoint to be analyzed.

5. A visually assisted zircon particle size refinement detection method according to claim 3, characterized in that: The specific method of obtaining the blocked pixel points and the prediction accuracy of the blocked pixel points according to the predicted pixel point sequences at the two endpoints includes: The two endpoints of the stack boundary line to be analyzed are recorded as the forward endpoint and the reverse endpoint respectively; The reverse sequence of the predicted pixel point sequence at the reverse endpoint is recorded as the reverse predicted pixel point sequence at the reverse endpoint; The two pixels with the same sequence number in the predicted pixel sequence at the forward endpoint and the reverse predicted pixel sequence at the reverse endpoint are recorded as a pixel pair; Each pixel pair is recorded as a pixel pair to be analyzed; The coordinate center points of the two pixels in the pixel pair to be analyzed are recorded as the blocked pixels; The Euclidean distance between two pixels in the pixel pair to be analyzed is recorded as the distance of the pixel pair to be analyzed; The sum of the distances of all pixel pairs is recorded as the prediction accuracy of the occluded pixel.

6. A visually assisted zircon particle size refinement detection method according to claim 3, characterized in that: The specific method of obtaining the local occlusion degree according to the prediction accuracy of the occluded pixel, the distance between the occluded pixel and the belonging boundary line, and the distance between the pixel on the stacking boundary line and the belonging boundary line is as follows: The sum of the shortest distances from all blocked pixels to their belonging boundary lines is recorded as the predicted partial blocking area; The sum of the shortest distances from all pixels on the stacking boundary line to the belonging boundary line is recorded as the fixed partial occlusion area; The difference between the number 1 and the power with the natural constant as the base and the opposite number of the prediction accuracy as the exponent is recorded as the occlusion weight; The product of the prediction accuracy and the fixed partial occlusion area is recorded as the fixed occlusion degree; The product of the predicted partial occlusion area and the occlusion weight is recorded as the predicted occlusion degree; The sum of the fixed occlusion degree and the predicted occlusion degree is recorded as the local occlusion degree of the stack boundary line to be analyzed.

7. A visually assisted zircon particle size refinement detection method according to claim 3, characterized in that: The specific method of obtaining the particle size characteristic confidence of the zircon particles according to the boundary line membership of the stacked boundary line, the local occlusion degree and the number of pixels inside the zircon particles, and then obtaining the effective zircon particles includes: The difference between the number 1 and the degree of membership of the boundary line is recorded as the occlusion judgment value of the stacked boundary line to be analyzed; The product of the local occlusion degree of the stack boundary line to be analyzed and the occlusion judgment value is recorded as the overall occlusion degree; The ratio of the sum of the overall occlusion degrees of all stacked boundary lines to the number of pixel points inside the zircon grain to be analyzed is recorded as the occlusion ratio of the zircon grain to be analyzed; The minimum value of the shielding ratio and the number 1 is recorded as the shielding characteristic value of the zircon grain to be analyzed; The difference between the number 1 and the obstruction characteristic value is taken as the particle size characteristic confidence of the zircon particles to be analyzed; Zircon particles whose particle size characteristic confidence is not equal to 0 are recorded as valid zircon particles.

8. A visually assisted zircon particle size refinement detection method according to claim 1, characterized in that: The specific method for obtaining the effective zircon particle size is as follows: The effective zircon particles are tested using the oriented maximum intercept method to obtain the oriented maximum intercept of the effective zircon particles in each direction; The maximum value of the maximum oriented intercept of the effective zircon particles in each direction is taken as the collected particle size of the effective zircon particles.

9. A visually assisted zircon particle size refinement detection method according to claim 1, characterized in that: The specific method of obtaining the particle size refinement index of the zircon sample according to the collected particle size of the effective zircon particles includes: The product of the mean of the collected particle sizes of all effective zircon particles and the standard deviation of the collected particle sizes of all effective zircon particles is recorded as the zircon particle size difference; The power of the natural constant as the base and the inverse number of the zircon particle size difference as the exponent is recorded as the particle size refinement index of the zircon sample.

10. A visually assisted zircon particle size refinement detection method according to claim 1, characterized in that: The specific method of obtaining the particle size refinement grade of the zircon sample according to the particle size refinement index includes: Classifying the particle size refinement level of the zircon sample whose particle size refinement index is less than or equal to the first preset threshold as low; Classifying the grain refinement level of the zircon sample whose grain refinement index is between the first preset threshold and the second preset threshold as medium; The grain refinement level of the zircon sample having a grain refinement index greater than or equal to the second preset threshold is classified as high.