Method for processing production image of mixed fertilizer

Abstract

The invention discloses a mixed fertilizer production image processing method, and belongs to the technical field of image monitoring, and the method comprises the following steps: S1, collecting a finished product image on a conveying belt, and constructing a uniform value for the finished product image; s2, determining a local result of each angular point according to the uniform value of the finished product image; and S3, according to the local result of each angular point in the finished product image, determining a suspected caking area, screening the suspected caking area, and determining a final caking area. According to the method, the uniformity degree of the whole image is quantified by constructing the uniformity value of the image instead of depending on a fixed threshold value; the uniform value suppresses the influence of local noise or sensor fluctuation based on global statistics (such as extreme values of 8 neighborhood RGB values); meanwhile, local analysis is only carried out on the angular points, the calculation amount is greatly reduced, meanwhile, key caking edges are reserved, isolated noise is eliminated through neighborhood connectivity analysis of the angular points, and real caking is reserved.

Description

Technical Field

[0001] The present invention belongs to the technical field of image monitoring, and particularly relates to an image processing method for the production of compound fertilizers. Background Art

[0002] In the field of chemical fertilizer production, the caking problem of compound fertilizers directly affects the product quality and usability. The caking phenomenon is usually caused by factors such as the hygroscopicity of raw materials, fluctuations in production process parameters, or insufficient cooling. After its formation, it will cause fertilizer particles to adhere into lumps, resulting in poor feeding and uneven dosage during application, and even triggering production accidents such as equipment blockage. Therefore, in the production process of compound fertilizers, real-time caking detection of the finished products on the conveyor belt is a key link to ensure product quality.

[0003] Traditional caking detection methods mainly rely on manual sampling inspection or simple mechanical screening, and have the following limitations: 1. Subjectivity and lag: Manual detection has low efficiency, is difficult to cover the entire process, and the results are affected by personnel experience, and real-time feedback cannot be achieved; 2. Destructive detection: Mechanical screening requires sampling for off-line analysis, which may damage the product integrity and cannot reflect the real-time state in the dynamic production of the conveyor belt; 3. Poor environmental adaptability: There are problems such as dust, vibration, and uneven illumination in the chemical fertilizer production site. Traditional visual detection methods are easily interfered, resulting in false detection or missed detection.

[0004] With the development of machine vision technology, automated detection based on image processing has become a research hotspot. However, the existing technologies still face challenges: 1. Difficult to extract caking features: The surface texture of compound fertilizer particles is complex, and the gray level and texture differences between the caking area and normal particles may change dynamically due to fluctuations in raw material ratios. Traditional global threshold segmentation or fixed template matching methods have insufficient robustness; 2. Conflict between computational efficiency and accuracy: High-precision algorithms (such as deep learning) require a large amount of labeled data and have poor real-time performance, while lightweight algorithms are easily interfered by noise and it is difficult to balance the detection speed and accuracy; 3. Weak adaptability to dynamic scenarios: In scenarios such as conveyor belt vibration, illumination change, and particle overlapping and stacking, traditional corner detection or edge analysis methods are prone to generate false targets. Summary of the Invention

[0005] In order to solve the above problems, the present invention proposes an image processing method for the production of compound fertilizers.

[0006] The technical solution of the present invention is: An image processing method for the production of compound fertilizers includes the following steps:

[0007] S1. Collect the image of the finished product on the conveyor belt and construct a uniformity value for the image of the finished product;

[0008] S2. Determine the local results of each corner point according to the uniformity value of the image of the finished product;

[0009] S3. Determine the suspected caking area based on the local results of each corner point in the finished product image, and screen the suspected caking area to determine the final caking area.

[0010] Further, S1 includes the following sub-steps:

[0011] S11. Collect the finished product image on the conveyor belt;

[0012] S12. Generate a two-dimensional matrix for each pixel position of the finished product image;

[0013] S13. Generate the uniformity value of the finished product image according to the two-dimensional matrices of all pixel positions.

[0014] The beneficial effect of the above further solution is: In the present invention, by constructing a two-dimensional matrix of pixel positions, the local neighborhood information (the L of the 8-neighborhood max , L min and the current pixel value) is encoded as matrix elements, and the local texture or edge information is transformed into a matrix form. By extracting the eigenvalues of the two-dimensional matrix and calculating the covariance, the local structural information is transformed into a global uniformity index; the covariance is used as the global uniformity value, which can comprehensively evaluate the uniformity of the entire image, rather than relying on local thresholds, and adapts to the dynamic changes of the surface texture of the compound fertilizer (such as fluctuations in raw material ratios).

[0015] Further, in S12, the expression of the two-dimensional matrix A of the pixel position is: , , , , ; where a 11 represents the first value of the two-dimensional matrix, a 22 represents the second value of the two-dimensional matrix, a 12 represents the third value of the two-dimensional matrix, a 21 represents the fourth value of the two-dimensional matrix, L max represents the maximum RGB value of the 8-neighborhood of the pixel position, L min represents the minimum RGB value of the 8-neighborhood of the pixel position, and l represents the RGB value of the pixel position.

[0016] Further, S13 includes the following sub-steps:

[0017] S131. Take the first eigenvalue of the two-dimensional matrices corresponding to all pixel positions as the first eigenvariable;

[0018] S132. Take the second eigenvalue of the two-dimensional matrices corresponding to all pixel positions as the second eigenvariable;

[0019] S133. Calculate the covariance between the first feature variable and the second feature variable as the uniformity value of the finished product image.

[0020] The beneficial effect of the above further solution is: in the present invention, the traditional method may misjudge the natural concavities and convexities of particles as caking, while covariance analysis effectively distinguishes true caking from surface texture changes by quantifying structural consistency.

[0021] Further, S2 includes the following sub-steps:

[0022] S21. Determine the weight coefficient of the pixel position using the uniformity value of the finished product image;

[0023] S22. Take the difference between 1 and the weight coefficient as the autocorrelation coefficient of the pixel position;

[0024] S23. Multiply the RGB values of each corner point in the finished product image by the autocorrelation coefficient as the local result of the corner point.

[0025] The beneficial effect of the above further solution is: in the present invention, the weight coefficient is determined by the global uniformity value (covariance), enabling the algorithm to adaptively focus on non-uniform regions (possibly caking). Only local results are calculated for corner points (significant points in the image, such as edge intersections), reducing the computational amount while retaining key structural information. Corner points are usually located at the edges of caking or texture mutation points. By analyzing the local results, the caking boundary can be accurately located. The joint judgment of structural consistency (autocorrelation coefficient) and color information avoids misjudging regions with abnormal colors but uniform structures as caking.

[0026] Further, in S21, the calculation formula for the weight coefficient k0 is: ; where C represents the uniformity value, e represents the exponent, and G represents the gradient magnitude of the pixel position.

[0027] The beneficial effect of the above further solution is: in the present invention, k0 is dynamically determined by the ratio of C to G, adapting to different production batches (such as raw material ratio, humidity change) and lighting conditions.

[0028] Further, S3 includes the following sub-steps:

[0029] S31. Take the corner points with local results less than the mean of all local results as the suspected caking regions;

[0030] S32. In the suspected caking regions, if there are pixel positions in the four-neighborhood of a corner point that belong to the suspected caking regions, then the corner point is determined as a caking point, otherwise the corner point is a normal region;

[0031] S33. Take all caking points as the final caking regions.

[0032] The beneficial effects of the above further solution are as follows: In the present invention, through the connectivity check of the four neighborhoods (up, down, left, and right), it is required that the caking area be spatially continuous. Isolated noise points are excluded because there are no other suspected caking points in their neighborhoods, significantly reducing false detections. The discrete caking points are merged into a continuous area without presetting the number of neighborhoods or distance thresholds, and it is completely based on data-driven connectivity judgment, adapting to cakings of different sizes.

[0033] The beneficial effects of the present invention are as follows: The present invention constructs the uniformity value of the image to quantify the uniformity degree of the entire image, rather than relying on a fixed threshold; the uniformity value is based on global statistics (such as the extreme values of the 8-neighborhood RGB values), suppressing the influence of local noise or sensor fluctuations; at the same time, the present invention only performs local analysis on the corner points, greatly reducing the amount of calculation, while retaining the key caking edges, excluding isolated noise through the neighborhood connectivity analysis of the corner points, and retaining the real cakings. The accurate positioning of the caking area can provide data support for the adjustment of process parameters (such as cooling time, raw material ratio), reducing the generation of cakings. Description of the Drawings

[0034] Figure 1 It is a flowchart of the production image processing method for the mixed fertilizer. Detailed Embodiments

[0035] The embodiments of the present invention will be further described below with reference to the drawings.

[0036] As Figure 1 shown, the present invention provides a production image processing method for the mixed fertilizer, including the following steps:

[0037] S1. Collect the finished product image on the conveyor belt and construct the uniformity value for the finished product image;

[0038] S2. Determine the local results of each corner point according to the uniformity value of the finished product image;

[0039] S3. Determine the suspected caking area according to the local results of each corner point in the finished product image, and screen the suspected caking area to determine the final caking area.

[0040] In the embodiment of the present invention, S1 includes the following sub-steps:

[0041] S11. Collect the finished product image on the conveyor belt;

[0042] S12. Generate a two-dimensional matrix for each pixel position of the finished product image;

[0043] S13. Generate the uniformity value of the finished product image according to the two-dimensional matrix of all pixel positions.

[0044] In the present invention, by constructing the two-dimensional matrix of the pixel positions, the local neighborhood information (the L of the 8-neighborhood)max , L min and the current pixel value) are encoded as matrix elements, converting the local texture or edge information into matrix form. By extracting the eigenvalues of the two-dimensional matrix and calculating the covariance, the local structural information is converted into a global uniformity index; the covariance, as the global uniformity value, can comprehensively evaluate the uniformity of the entire image, rather than relying on local thresholds, adapting to the dynamic changes in the surface texture of the compound fertilizer (such as fluctuations in raw material ratios).

[0045] In the embodiment of the present invention, in S12, the expression of the two-dimensional matrix A at the pixel position is: , , , , ; where a 11 represents the first value of the two-dimensional matrix, a 22 represents the second value of the two-dimensional matrix, a 12 represents the third value of the two-dimensional matrix, a 21 represents the fourth value of the two-dimensional matrix, L max represents the maximum RGB value of the 8-neighborhood at the pixel position, L min represents the minimum RGB value of the 8-neighborhood at the pixel position, and l represents the RGB value at the pixel position.

[0046] In the embodiment of the present invention, S13 includes the following sub-steps:

[0047] S131. Take the first eigenvalue of the two-dimensional matrix corresponding to all pixel positions as the first feature variable;

[0048] S132. Take the second eigenvalue of the two-dimensional matrix corresponding to all pixel positions as the second feature variable;

[0049] S133. Calculate the covariance between the first feature variable and the second feature variable as the uniformity value of the finished image.

[0050] In the present invention, the traditional method may misjudge the natural concavities and convexities of particles as caking, while covariance analysis effectively distinguishes true caking from surface texture changes by quantifying structural consistency.

[0051] In the embodiment of the present invention, S2 includes the following sub-steps:

[0052] S21. Determine the weight coefficient at the pixel position using the uniformity value of the finished image;

[0053] S22. Take the difference obtained by subtracting the weight coefficient from 1 as the autocorrelation coefficient at the pixel position;

[0054] S23. Multiply the RGB value of each corner point in the finished image by the autocorrelation coefficient as the local result of the corner point.

[0055] In the present invention, the weight coefficient is determined by the global uniformity value (covariance), enabling the algorithm to adaptively focus on non-uniform regions (which may be lumps). Only local results are calculated for corner points (significant points in the image, such as edge intersections), reducing the computational amount while retaining key structural information. Corner points are usually located at the edges of lumps or texture mutation points, and the lump boundaries can be accurately located through local result analysis. The joint judgment of structural consistency (autocorrelation coefficient) and color information avoids misjudging regions with abnormal colors but uniform structures as lumps.

[0056] In the embodiment of the present invention, in S21, the calculation formula for the weight coefficient k0 is: ; where C represents the uniformity value, e represents the exponent, and G represents the gradient magnitude of the pixel position.

[0057] The beneficial effect of the above further solution is that in the present invention, k0 is dynamically determined by the ratio of C to G, adapting to different production batches (such as raw material ratios, humidity changes) and lighting conditions.

[0058] In the embodiment of the present invention, S3 includes the following sub-steps:

[0059] S31. Take the corner points where the local results are less than the average of all local results as suspected lump regions;

[0060] S32. In the suspected lump regions, if there are pixel positions in the four-neighborhood of a corner point that belong to the suspected lump regions, then this corner point is determined as a lump point; otherwise, this corner point is a normal region;

[0061] S33. Take all lump points as the final lump regions.

[0062] In the present invention, through the connectivity check of the four-neighborhood (up, down, left, and right), it is required that the lump regions are spatially continuous. Isolated noise points are excluded because there are no other suspected lump points in their neighborhoods, significantly reducing false detections. The discrete lump points are merged into continuous regions without presetting the number of neighborhoods or distance thresholds, and it is completely based on data-driven connectivity judgment, adapting to lumps of different sizes.

[0063] Those of ordinary skill in the art will realize that the embodiments described herein are for helping readers understand the principles of the present invention, and it should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those of ordinary skill in the art can make various other specific deformations and combinations that do not depart from the essence of the present invention based on the technical revelations disclosed in the present invention, and these deformations and combinations are still within the protection scope of the present invention.

Claims

1. A method for processing images in the production of compound fertilizers, characterized in that, Including the following steps: S1. Collect the finished product images on the conveyor belt and construct a uniformity value for the finished product images; S2. Determine the local results of each corner point according to the uniformity value of the finished product images; S3. Determine the suspected caking area according to the local results of each corner point in the finished product images, and screen the suspected caking area to determine the final caking area.

2. The method for processing images in the production of compound fertilizers according to claim 1, characterized in that, The S1 includes the following sub-steps: S11. Collect the finished product images on the conveyor belt; S12. Generate a two-dimensional matrix for each pixel position of the finished product images; S13. Generate the uniformity value of the finished product images according to the two-dimensional matrices of all pixel positions.

3. The method for processing images in the production of compound fertilizers according to claim 2, characterized in that, In S12, the expression of the two-dimensional matrix A of pixel positions is: , , , , ; where a 11 represents the first value of the two-dimensional matrix, a 22 represents the second value of the two-dimensional matrix, a 12 represents the third value of the two-dimensional matrix, a 21 represents the fourth value of the two-dimensional matrix, L max represents the maximum RGB value of the 8-neighborhood of the pixel position, L min represents the minimum RGB value of the 8-neighborhood of the pixel position, and l represents the RGB value of the pixel position.

4. The method for processing images in the production of compound fertilizers according to claim 2, characterized in that, The S13 includes the following sub-steps: S131. Take the first eigenvalue of the two-dimensional matrix corresponding to all pixel positions as the first eigenvariable; S132. Take the second eigenvalue of the two-dimensional matrix corresponding to all pixel positions as the second eigenvariable; S133. Calculate the covariance between the first eigenvariable and the second eigenvariable as the uniformity value of the finished product images.

5. The method for processing images in the production of compound fertilizers according to claim 1, characterized in that, The S2 includes the following sub-steps: S21. Determine the weight coefficient of the pixel position by using the uniformity value of the finished product images; S22. Take the difference between 1 and the weight coefficient as the autocorrelation coefficient of the pixel position; S23. Take the product of the RGB values of each corner point in the finished product images and the autocorrelation coefficient as the local result of the corner point.

6. The method for processing images in the production of compound fertilizers according to claim 5, characterized in that, In the above S21, the calculation formula of the weight coefficient k0 is as follows: ; where C represents the uniform value, e represents the exponent, and G represents the gradient amplitude of the pixel position.

7. The method for processing images in the production of compound fertilizers according to claim 5, characterized in that, The S3 includes the following sub-steps: S31. Take the corner points with local results less than the mean of all local results as the suspected caking area; S32. In the suspected caking area, if there is a pixel position in the four-neighborhood of the corner point that belongs to the suspected caking area, then the corner point is determined as a caking point, otherwise the corner point is a normal area; S33. Take all caking points as the final caking area.

Images