A visual determination method and system for dissolution rate of water-soluble fertilizer

Abstract

The present application relates to the technical field of image data processing, and especially relates to a water-soluble fertilizer dissolution rate visual determination method and system, which comprises the following steps: acquiring sequence images of a dissolution process; performing multi-scale frequency domain analysis on the images, calculating a scale convergence index by using the deviation of local phase and weighted average phase; calculating a spectral sharpness factor according to the ratio of high-frequency and low-frequency energy response; fusing the scale convergence index and the spectral sharpness factor to obtain a solid boundary probability; based on the probability, segmenting a particle region and converting it into physical mass, and differentiating the mass change sequence to obtain a dissolution rate. By fusing phase consistency and frequency energy distribution characteristics, the present application effectively distinguishes physical entity edges and optical refraction halos, solves the refraction interference problem during high-concentration dissolution, and improves the accuracy of dissolution rate determination.

Description

Technical Field

[0001] This invention relates to the field of image data processing technology, and in particular to a method and system for visually measuring the dissolution rate of water-soluble fertilizer. Background Technology

[0002] Water-soluble fertilizers, as the core carrier of fertigation technology, occupy an important position in agricultural production, and their dissolution rate is one of the key physical indicators for evaluating their product quality. Poor fertilizer solubility can easily lead to clogging of drip irrigation nozzles during application, affecting the uniformity of water and fertilizer delivery and the efficiency of crop nutrient absorption. Therefore, monitoring and determining the dissolution process and rate of water-soluble fertilizers in a liquid environment is of great significance.

[0003] Existing technologies typically employ machine vision systems combined with image processing algorithms to dynamically monitor the dissolution behavior of fertilizer granules. These methods acquire sequential images of the dissolution process using industrial cameras and extract fertilizer granule contour features using edge detection operators or phase consistency algorithms. The remaining projected area of ​​the fertilizer granules is then calculated to infer the dissolution rate. Among these methods, the phase consistency algorithm, due to its robustness to changes in image brightness and contrast, is often used to overcome detection interference caused by uneven lighting and to obtain fertilizer granule morphology data.

[0004] However, existing technologies have limitations when measuring high-concentration water-soluble fertilizers rich in mineral salts: During dissolution, water-soluble fertilizers form a high-concentration diffusion layer around the solid-liquid interface, causing refracted halos when light passes through this area due to changes in the refractive index gradient. Existing feature extraction algorithms, such as those for phase consistency, typically possess scale invariance; that is, as long as structural features exist locally in the image, whether these features are physically sharp solid edges or optically broad refracted halo bands, the algorithm will classify them as structural information with high response values. Furthermore, the algorithms fail to consider the fundamental differences in frequency domain energy distribution between solid-liquid interfaces and concentrated / dilute liquid interfaces, making it difficult to distinguish between physical abrupt changes and optical gradual transitions. This leads to the misjudgment of halos as undissolved fertilizer particle boundaries when processing dissolution images with strong refracted halos, causing the calculated dissolution rate to deviate from the true value and failing to meet the requirements for high-precision detection. Summary of the Invention

[0005] To address the technical problem that existing feature extraction algorithms, limited by scale invariance, cannot distinguish between refracted halo interference and the actual fertilizer particle boundary, thus leading to inaccurate dissolution rate measurements, this invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides a method for visually measuring the dissolution rate of water-soluble fertilizer, the method comprising the steps of:

[0007] A sequence of images of fertilizer granules during the dissolution process is acquired. Using any image in the sequence as the target image, multi-scale frequency domain analysis is performed on each pixel in the target image to obtain the local phase and energy components at each scale. A weighted average phase is calculated using the energy components. Based on the phase deviation between the local phase and the weighted average phase, the scale convergence index of each pixel is calculated. All scales are divided into a first scale group and a second scale group. The energy response of each pixel in the target image is obtained under the first and second scale groups. The spectral sharpness factor of each pixel is calculated based on the ratio of the energy response of the first scale group to that of the second scale group. The scale convergence index and the spectral sharpness factor are fused to obtain the solid boundary probability of each pixel. The target region of undissolved granules is segmented based on the solid boundary probability. The physical mass of the fertilizer is calculated based on the area of ​​the target region. Each image in the sequence is processed sequentially to obtain a sequence of physical mass changes over time. The sequence of changes is differentiated to calculate the dissolution rate of the fertilizer.

[0008] This invention acquires sequential images of the water-soluble fertilizer dissolution process and performs multi-scale frequency domain analysis. It obtains a scale convergence index by measuring the deviation between the local phase and the weighted average phase at different scales, thereby evaluating the phase consistency of pixels. Simultaneously, the frequency domain energy is divided into a first-scale group and a second-scale group, and the ratio of their energy responses is used to calculate a spectral sharpness factor, thus assessing edge sharpness. Furthermore, by fusing these two indices, the probability of solid boundaries is obtained. This approach combines the adaptability of phase consistency to changes in illumination with the ability of frequency domain energy distribution to distinguish between physical edges and optical ghosting. Even with the interference of high-concentration refractive halos generated during water-soluble fertilizer dissolution, it reduces the misclassification of optically refracted regions as solid particles, thus more accurately segmenting the target area of ​​undissolved particles and calculating their physical mass. Finally, by differentiating the physical mass sequence, a continuous dissolution rate consistent with the actual physical process is obtained.

[0009] Preferably, the scale convergence index of the pixel is... Satisfying the relation:

[0010] ;

[0011] in, It is a pixel. In the Energy amplitude at various scales; It is a pixel. In the Local phase at various scales; It is a pixel. The weighted average phase; It represents the total number of filter scales; It is the index value at the current scale; It is the preset first minute value; It is a refractive sensitivity control factor; It is a pixel. The standard deviation of the phase difference between the local phase and the weighted average phase at each scale; It is the hyperbolic tangent function.

[0012] This invention utilizes a relationship involving cosine and hyperbolic tangent functions to calculate the scale convergence index. The energy amplitude-weighted cosine term reflects the directional strength of phase convergence at each scale, while the hyperbolic tangent term based on the standard deviation of the phase difference serves as a suppression factor. When the phase dispersion of a pixel is large across different scales, this suppression factor reduces the calculated value. This calculation mechanism can preserve the true physical edge signal while suppressing refracted halo regions with strong energy but diverging phases with scale, thereby improving the accuracy of feature extraction.

[0013] Preferably, the spectral sharpness factor of the pixel Satisfying the relation:

[0014] ;

[0015] in, It is a pixel. In the Energy amplitude at various scales; It is the sharpness sensitivity coefficient; It is the boundary index value between the first scale group and the second scale group; It is the preset second minute value; It is a natural exponential function.

[0016] This invention uses a nonlinear relation based on the natural exponential function to calculate the spectral sharpness factor. By analyzing the ratio of the energy sum of the high-frequency scale group to the energy sum of the low-frequency scale group, it reflects the spectral attenuation characteristics of the local image. Since the edges of physical entities usually have rich high-frequency components, while the halo formed by optical diffusion is mainly concentrated in the low-frequency part, this relation can map this difference in frequency domain distribution into a numerical sharpness index. Through the mapping effect of the natural exponential function, it further enhances the distinguishability of blurred areas lacking high-frequency details.

[0017] Preferably, the step of fusing the scale convergence index and the spectral sharpness factor to obtain the solid boundary probability of each pixel includes: introducing a preset fusion index to exponentially weight the spectral sharpness factor to obtain a weighted spectral sharpness factor; and recording the product of the scale convergence index and the weighted spectral sharpness factor as the solid boundary probability of the pixel.

[0018] This invention introduces a preset fusion index to nonlinearly weight the spectral sharpness factor, and multiplies the weighted result with the scale convergence index to obtain the solid boundary probability. This multiplication fusion strategy is equivalent to establishing a logical AND relationship, requiring that the pixels judged as solid boundaries must have both high phase consistency and high spectral sharpness, thereby further reducing the false detection rate that may be caused by single feature judgment and ensuring that only the region that best matches the physical entity characteristics is retained.

[0019] Preferably, the step of calculating the physical mass of the fertilizer based on the area of ​​the target region includes: using an adaptive threshold segmentation algorithm to binarize the solid boundary probability map composed of the solid boundary probabilities of all pixels in the target image to obtain a fertilizer particle mask; counting the total number of pixels within the fertilizer particle mask; and converting the total number of pixels into physical size based on camera calibration parameters, and calculating the physical mass of the fertilizer particles based on an equivalent sphere model.

[0020] Preferably, the fertilizer has a dissolution rate Satisfying the relation:

[0021] ;

[0022] in, It is about time Differentiation; It refers to the density of the fertilizer; yes The total number of pixels in the fertilizer particle mask in the time-series image; It is pi; It is the camera's pixel equivalent.

[0023] This invention performs time-dependent derivative operations on the physical mass change sequence obtained based on an equivalent sphere model. By introducing physical parameters such as density, pi, and pixel equivalent, discrete mass change data is transformed into a continuous instantaneous rate curve. This differential processing method can capture the dynamic fluctuations of the rate during the dissolution process. Compared with simply calculating the average dissolution time, it can more accurately reflect the speed characteristics of fertilizer at different dissolution stages.

[0024] Preferably, the adaptive threshold segmentation algorithm is the Otsu algorithm.

[0025] Preferably, the calculation of the fertilizer's dissolution rate includes: comparing the remaining physical mass of the fertilizer with a preset initial mass to obtain a physical mass percentage; defining the moment when the physical mass percentage first falls below a preset residue threshold as the complete dissolution time; and determining that the current fertilizer's dissolution performance is qualified if the complete dissolution time is less than a preset standard threshold.

[0026] Preferably, the multi-scale frequency domain analysis employs a logarithmic Gabor filter bank for multi-scale decomposition.

[0027] In a second aspect, the present invention provides a visual measurement system for the dissolution rate of water-soluble fertilizer. The visual measurement system for the dissolution rate of water-soluble fertilizer includes a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, the visual measurement method for the dissolution rate of water-soluble fertilizer according to the first aspect of the present invention is implemented.

[0028] By adopting the above technical solution, a computer program for visually measuring the dissolution rate of water-soluble fertilizer according to the first aspect of the present invention is generated and stored in a memory so that it can be loaded and executed by a processor. A terminal device can then be made based on the memory and the processor for convenient use.

[0029] The beneficial effects of this invention are as follows: This invention addresses the problem of inaccurate visual detection caused by concentration refraction halos during the dissolution process of water-soluble fertilizers. By integrating phase consistency analysis and high-low frequency energy ratio analysis in the multi-scale frequency domain, it evaluates image features from two dimensions: phase stability and spectral distribution. Utilizing the essential difference in frequency domain response between physical entity edges and optical refraction ghost images, it reduces the impact of halo interference on particle contour extraction, improving the accuracy of solid particle detection in complex liquid environments. This invention uses an equivalent sphere model and camera calibration parameters to map the two-dimensional image projection area to a three-dimensional physical mass. It calculates the instantaneous dissolution rate through differential processing of the time series. This method not only focuses on the endpoint of dissolution but also on the dynamic changes in the dissolution process, providing a more physically accurate data dimension for evaluating fertilizer dissolution performance. This invention combines adaptive threshold segmentation with objective evaluation indicators. By automatically determining the segmentation threshold and judging the complete dissolution time based on the residue threshold, it reduces the subjectivity of manual operation and the uncertainty caused by environmental changes. It not only achieves the measurement of dissolution rate but also automatically judges product qualification according to preset standards, making it suitable for quality monitoring in the research and development and production process of water-soluble fertilizers. Attached Figure Description

[0030] Figure 1 A flowchart of a method for visually measuring the dissolution rate of water-soluble fertilizer provided in an embodiment of the present invention;

[0031] Figure 2 A comparison chart of water-soluble fertilizer dissolution rates calculated by different methods provided in the embodiments of the present invention;

[0032] Figure 3 This is a structural block diagram of a visual measurement system for the dissolution rate of water-soluble fertilizer provided in an embodiment of the present invention. Detailed Implementation

[0033] The first aspect of this invention provides a method for visually measuring the dissolution rate of water-soluble fertilizer, such as... Figure 1 As shown, the method includes steps S100-S400:

[0034] Step S100: Obtain sequence images of fertilizer granules during the dissolution process.

[0035] It should be noted that, in order to accurately determine the solubility characteristics of water-soluble fertilizers in a liquid environment, high-quality image data is required as the basis for analysis. This invention mainly targets fully water-soluble fertilizers used in precision agriculture, such as high-end water-soluble fertilizers containing mineral salts or amino acids. These fertilizers are prone to generating refracted halos around fertilizer particles during dissolution due to concentration gradients, affecting measurement accuracy. Therefore, this step requires constructing an optical acquisition environment capable of clearly capturing changes in the morphology of fertilizer particles.

[0036] Specifically, in a pre-set dissolution experimental environment, an image acquisition component is deployed to monitor fertilizer particles in real time. To ensure that the image features can truly reflect the physical form, it is preferable to use an industrial camera with a telecentric lens and a backlit parallel light source for shooting. The backlight illumination makes the fertilizer particles appear as high-contrast black silhouettes, while the refracted halo appears as gray semi-transparent. Under these optical conditions, multiple images are continuously acquired at a pre-set sampling frequency, and the multiple images are arranged in chronological order of acquisition time to obtain a sequence of images containing the entire dissolution process.

[0037] Thus, the sequence images under the dissolution environment were obtained.

[0038] Step S200: Taking any image in the sequence of images as the target image, perform multi-scale frequency domain analysis on each pixel in the target image to obtain the local phase and energy components at each scale, calculate the weighted average phase using the energy components, and calculate the scale convergence index of each pixel based on the phase deviation between the local phase and the weighted average phase.

[0039] It should be noted that while existing phase consistency algorithms can reduce the interference of uneven illumination when extracting image features, they exhibit scale invariance. This means that regardless of whether there are sharp solid edges or wide refractive halo bands, as long as structural features exist, the algorithm will assign a high response value. To distinguish between the two, this invention utilizes the fact that the solid-liquid interface is a physically abrupt change, with its phase being highly consistent across different frequency scales; whereas the concentrated-dilute liquid interface is an optically gradual change, and its phase drifts with frequency scale variations. Therefore, this step introduces a scale convergence index to eliminate the interference of refractive halos by suppressing pixels whose phase changes drastically with scale.

[0040] Specifically, firstly, for any pixel in the image sequence, logarithmic Gabor wavelet transform is preferably used for multi-scale decomposition. Regarding the configuration of the number of filter banks and frequency coverage, as a preferred implementation, the number of scales can be set to... One, and the wavelength range covers Pixels. It should be noted that the above parameters are set based on the following: the refracted halos produced by the dissolution of water-soluble fertilizers typically have a wide frequency distribution range. If the scale is too small, such as only 2 pixels, it is difficult to capture the continuous drift trend of phase change with frequency, leading to suppression failure; if the scale is too large, such as more than 10 pixels, it may introduce too much high-frequency noise and significantly increase the computational burden; while a wavelength range of 3 to 30 pixels can completely cover the morphological features from fine edges to wide halos. Implementers can fine-tune within the above range according to the camera's pixel resolution and the expected minimum halo width.

[0041] Then, the local phase and weighted average phase of each pixel are calculated. It should be noted that obtaining these two intermediate variables is a prerequisite for evaluating the stability of phase behavior. The local phase reflects the local structural information of the image at a specific frequency scale; the weighted average phase uses energy weighting to extract the dominant structural orientation under multi-scale consensus.

[0042] Specifically, for any scale, the real and imaginary parts of the response are obtained through convolution operation between the filter and the image, and the local phase at that scale is calculated using the arctangent function. The phases at each scale are treated as vector directions and the corresponding energy amplitudes are treated as vector lengths, and vector superposition is performed. The angle of the superimposed vector is calculated to obtain the weighted average phase, which represents the dominant direction of the image structure. The angle difference between the local phases at each scale and the weighted average phase is statistically analyzed, and the standard deviation of this difference is calculated to obtain the phase dispersion.

[0043] Next, the scale convergence index of each pixel is calculated. It should be noted that, based on physical optics properties, the boundary of fertilizer particles is a step signal, and its phase direction is highly consistent across all frequency bands; while the refracted halo is a gradual signal, and its phase direction shifts with scale. To distinguish between solid objects and halos, it is necessary not only to consider the energy superposition intensity of pixels at various scales, but also to consider their phase dispersion across scales, in order to eliminate optical artifacts that have strong energy but divergent phases.

[0044] Based on the above logic, pixels Scale convergence index Satisfying the relation:

[0045] ;

[0046] in, It is a pixel. In the Energy amplitude at various scales; It is a pixel. In the Local phase at various scales; It is a pixel. The weighted average phase; It represents the total number of filter scales; It is the index value at the current scale; It is a preset first tiny value used to prevent the denominator of the first part from being 0, and can be set to 0.001; It is a refractive sensitivity control factor; It is a pixel. The standard deviation of the phase difference between the local phase and the weighted average phase at each scale; It is the hyperbolic tangent function.

[0047] In this relation, the first part Used to measure phase consistency, the larger the value of this item, the better the pixel consistency. The phases at all scales tend to converge in the same direction, indicating that there is a clear image structure at this point; Part Two Used to measure phase stability, when pixel The greater the phase dispersion, the closer the hyperbolic tangent function value approaches 1, causing the second part to approach 0. The product of the two makes... High values ​​are only output when pixels satisfy both significant energy superposition and high phase convergence across scales, thus effectively screening out potential real fertilizer particle regions.

[0048] It should be noted that the refractive sensitivity control factor The settings need to be determined based on the type of water-soluble fertilizer. If the sample to be tested is an amino acid-containing product, the phase shift is more pronounced due to the high viscosity and wide halo of the solution during dissolution. Therefore, the settings can be adjusted accordingly. Set it to a smaller value, such as 0.5, to reduce the false rejection of weak true boundaries; if the sample to be tested is a pure inorganic salt product, the halo is narrow but the gradient is large, it should be set to a smaller value. Set it to a higher value, such as 1.0, to enhance the ability to eliminate false edges.

[0049] At this point, the scale convergence index of each pixel has been obtained.

[0050] Step S300: Divide all scales into a first scale group and a second scale group, obtain the energy response of each pixel in the target image under the first scale group and the second scale group, and calculate the spectral sharpness factor of each pixel according to the ratio of the energy response of the first scale group to the energy response of the second scale group.

[0051] It should be noted that the edges of physical entities are step signals, with their frequency domain energy distribution covering the entire frequency band and still exhibiting a strong response in the high-frequency range; while the edges formed by optical refraction are ramp signals, whose high-frequency energy decays rapidly. Therefore, this step introduces a spectral sharpness factor to further identify the physical properties of the edges by evaluating the ratio of high-frequency energy to low-frequency energy.

[0052] Specifically, the multi-scale filter bank is first divided into a first-scale group and a second-scale group. The boundary between the first-scale group and the second-scale group is usually half the total number of scales. It should be noted that this boundary strategy is based on the fact that refracted halos, as an optical diffusion phenomenon, have their energy mainly concentrated on the low-frequency, large-scale components, while the sharp boundaries of fertilizer particles can excite high-frequency responses at small scales. By comparing the energy intensity on both sides of the boundary line, the frequency domain attenuation characteristics of the signal can be most intuitively reflected.

[0053] For example, if the total number of filter scales If the value is 6, the first three smaller scales can be classified as the first scale group, and the last three larger scales can be classified as the second scale group.

[0054] Then, the spectral sharpness factor of each pixel is calculated. It should be noted that, in order to construct a probability index that takes values ​​in the range [0,1] and is highly sensitive to high-frequency components, this invention introduces a negative exponential function for nonlinear mapping and converts blurriness into sharpness through an inversion operation.

[0055] Based on the above logic, pixels Spectral sharpness factor Satisfying the relation:

[0056] ;

[0057] in, It is a pixel. spectral sharpness factor; It is a pixel. In the Energy amplitude at various scales; It is the sharpness sensitivity coefficient; It is the boundary index value between the first scale group and the second scale group; It is a preset second tiny value used to prevent the denominator in the fraction in the natural exponential function from being 0; it can be set to 0.001. It is a natural exponential function.

[0058] In this relation, the numerator of the fraction in the natural exponential function The sum of the energies of the first scale group, and the denominator The sum of the energies of the second scale group and the ratio of the sum are used to calculate the proportion of the energy of the first scale group relative to the energy of the second scale group. For refracted halos, due to optical diffusion, high frequencies are lost, the numerator is smaller than the denominator, causing the fractional value to approach 0; this, in turn, causes the natural exponential function value to approach 1. By subtracting the natural exponential function value from 1, the logic is reversed, resulting in... A value approaching 0 indicates that the pixel lacks the sharpness of a physical entity. For real fertilizer particles, their physical boundaries are sharp, and their high-frequency energy is abundant, resulting in a large fractional value in the natural exponential function; this, in turn, causes the natural exponential function value to approach 0, ultimately leading to... A value approaching 1 indicates that the pixel has high spectral sharpness.

[0059] It should be noted that the sharpness sensitivity coefficient... This is used to adjust the model's tolerance to high-frequency defects. If the overall optical blur is caused by turbidity in the shooting environment, the tolerance can be appropriately reduced. A value, such as 2.0, can relax the sharpness requirements and avoid missed detections; if the image clarity is extremely high, it should be increased. The value, such as 5.0, is used to rigorously filter the sharpest boundaries. In this embodiment, it is preferably set to 3.0 to achieve the best balance between the false negative rate and the false positive rate, adapting to most detection scenarios.

[0060] At this point, the spectral sharpness factor of each pixel has been obtained.

[0061] Step S400: Fuse the scale convergence index and spectral sharpness factor to obtain the solid boundary probability of each pixel; segment the target region of undissolved particles according to the solid boundary probability, and calculate the physical mass of the fertilizer based on the area of ​​the target region; process each image in the sequence image one by one to obtain the sequence of changes in physical mass over time, perform differential processing on the sequence of changes, and calculate the dissolution rate of the fertilizer.

[0062] It should be noted that the scale convergence index focuses on eliminating artifacts caused by phase instability, while the spectral sharpness factor focuses on eliminating blurred boundaries due to high-frequency missing data. To obtain a highly reliable fertilizer particle region, this step employs a nonlinear fusion strategy, combining the two to obtain the solid boundary probability. This accurately reconstructs the real-time physical morphology of the fertilizer particles, providing reliable data support for the subsequent accurate determination of the dissolution rate. Based on this, the evolution of the remaining mass over time is evaluated, ultimately achieving accurate determination of the dissolution rate of water-soluble fertilizer and an objective assessment of its dissolution performance.

[0063] Specifically, for each pixel in the image sequence, an exponential weighting mechanism is used to fuse the scale convergence index and the spectral sharpness factor. It should be noted that, to prevent regions with low spectral sharpness from interfering with the final determination, this invention introduces exponential modulation logic to non-linearly weight the spectral sharpness factor, thereby obtaining the confidence level that the pixel belongs to a real fertilizer particle.

[0064] Based on the above logic, pixels solid boundary probability Satisfying the relation:

[0065] ;

[0066] in, It is a pixel. The probability of solid boundaries; It is a pixel. The scale convergence index; It is a pixel. spectral sharpness factor; It is the integration index.

[0067] This relationship uses multiplicative logic, and only when a pixel simultaneously possesses a highly aligned phase... Larger values ​​and abundant high-frequency components When the value is large, its solid boundary probability Only then will it approach 1. (Index) This plays an exponential weighting role, controlling the suppression weight of the spectral sharpness factor on the final result; when A value greater than 1 enhances the suppressive effect of the sharpness factor on the result; that is, even a slight deficiency in sharpness will lower the final probability. In this embodiment, preferably... The value is set to 1.5 to ensure that only the most certain solid core is retained.

[0068] Secondly, all pixels in the sequence image are traversed, and their solid boundary probabilities are calculated respectively, thereby constructing a complete solid boundary probability map.

[0069] Next, the physical mass of the fertilizer granules is extracted based on the solid boundary probability map. It should be noted that the solid boundary probability map only represents the confidence level at the image level; to determine the dissolution rate, the image features need to be mapped to physical world quality indicators. Considering that the water-soluble fertilizer granules to be tested, such as urea and compound fertilizer granules, are usually prepared into approximately spherical shapes during the granulation process, and that in an isotropic static dissolution environment, fertilizer granules tend to maintain their original geometric characteristics and shrink uniformly, this invention uses an equivalent sphere model to establish a mathematical mapping relationship between the two-dimensional projected area and the three-dimensional physical mass.

[0070] Specifically, firstly, an adaptive threshold segmentation algorithm, such as Otsu, is used to binarize the solid boundary probability map, and regions with solid boundary probabilities greater than the threshold are marked as fertilizer particle masks; then, the total number of pixels within the mask region is counted; finally, based on the camera calibration parameters and the equivalent sphere model, the number of pixels is converted into the remaining physical mass of the fertilizer particles.

[0071] Then, the dissolution rate is obtained by differentiating the curve of the remaining physical mass over time. It should be noted that the dissolution of water-soluble fertilizer is a dynamic, non-linear process; obtaining only the remaining mass at a single moment cannot intuitively reflect the rate of dissolution. Therefore, this invention introduces calculus processing logic to obtain the instantaneous dissolution rate at each moment by calculating the first negative derivative of the remaining mass with respect to time.

[0072] Based on the above logic, the dissolution rate satisfies the following relationship:

[0073] ;

[0074] in, It is fertilizer Dissolution rate at time; It is about time Differential operators; The density of the fertilizer can be obtained through pre-measurement or by consulting the product specifications. yes The total number of pixels in the fertilizer particle mask in the time-series image; It is pi; It is the camera's pixel equivalent, measured in millimeters per pixel.

[0075] This relation is constructed based on an equivalent sphere model. The calculation logic is as follows: using the inverse function of the circle area relationship, the total number of pixels is... Convert to equivalent pixel radius, multiply by pixel equivalent Convert it to a physical millimeter radius; calculate the fertilizer particle volume using the spherical volume formula, and multiply by the density. get The remaining mass of the fertilizer granules at time is calculated; finally, the negative derivative of the remaining mass with respect to time is obtained. The rate of mass loss at any given time is also known as the dissolution rate.

[0076] It should be added that, regarding the acquisition of the camera's pixel equivalent: before conducting the dissolution experiment, a standard calibration plate with known physical dimensions, such as a checkerboard calibration plate, needs to be placed at the shooting plane position of the reaction vessel. By taking an image of the calibration plate and calculating the ratio of the pixel distance of the feature points of the calibration plate in the image to the actual physical distance, the physical size represented by a unit pixel, i.e., the pixel equivalent, can be obtained.

[0077] like Figure 2 The figure shows a comparison of the dissolution rates of water-soluble fertilizers calculated by different methods. The horizontal axis represents dissolution time, and the vertical axis represents dissolution rate. The dashed curve corresponds to the true dissolution rate, the solid curve with a circular marker corresponds to the dissolution rate calculated by the traditional method, and the solid curve with a star marker corresponds to the dissolution rate calculated by the method of this invention. Observing the curve trends in the figure: In the initial stage of dissolution, the curve corresponding to the traditional method deviates significantly from the true dissolution rate, indicating that the traditional method is affected by optical refraction halos, leading to misjudgments of particle size and dissolution rate. In the middle and later stages of dissolution, due to the persistent weak concentration diffusion layer around the particles, the traditional method still exhibits a certain systematic deviation and is difficult to completely revert to the true value. In contrast, the curve corresponding to the method of this invention maintains a consistent trend with the true dissolution rate curve overall. Even with slight fluctuations in the initial stage, its deviation from the true dissolution rate is smaller than that of the traditional method, and it exhibits better following throughout the process. This indicates that by introducing scale convergence and spectral sharpness features, this invention can reduce optical artifact interference and improve the accuracy of capturing the true dissolution rate.

[0078] Finally, based on the changes in the dissolution rate and remaining mass, the complete dissolution time of the water-soluble fertilizer is determined, and the product grade is accordingly identified. It should be noted that, to achieve an objective assessment of the full water solubility characteristics, key time points need to be extracted from the continuous rate curve as the basis for judgment.

[0079] Specifically, monitoring The ratio of the remaining mass to the initial mass at a given time is used to determine the time of complete dissolution. When this ratio first drops to a preset residue threshold, that time is marked as the complete dissolution time. Subsequently, this complete dissolution time is compared with a preset product quality standard threshold. If the complete dissolution time is less than the standard threshold, the dissolution rate of the water-soluble fertilizer is deemed acceptable; otherwise, it is deemed unacceptable.

[0080] It should be further explained that the residue threshold represents the stringency of the definition of complete dissolution. For drip irrigation systems with high requirements, this threshold should be set very low, such as 0.5% or lower, to ensure that tiny particles do not cause blockages. For conventional scenarios, it can be appropriately relaxed, such as setting it to 1.0%. In this embodiment, 0.5% is preferred to ensure high-end quality. The product quality standard threshold is the upper limit of time set based on the company's factory standard or industry standard. For example, if complete dissolution is required within 3 minutes, the threshold is 180 seconds. This value should be specifically set according to the physical characteristics and market positioning of different fertilizer formulations.

[0081] This concludes the determination of the dissolution rate of water-soluble fertilizer and the final evaluation of its dissolution performance.

[0082] The second aspect of this embodiment provides a visual measurement system for the dissolution rate of water-soluble fertilizer, such as... Figure 3 As shown, the visual measurement system for the dissolution rate of water-soluble fertilizer includes a memory and a processor. The memory stores computer program instructions, and when the computer program instructions are executed by the processor, a visual measurement method for the dissolution rate of water-soluble fertilizer according to the first aspect of the present invention is implemented.

[0083] The visual measurement system for the dissolution rate of water-soluble fertilizer also includes other components well known to those skilled in the art, such as communication buses and communication interfaces. Their setup and functions are known in the art and will not be described in detail here.

[0084] In this invention, the aforementioned memory can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, apparatus, or device. For example, a computer-readable storage medium can be any suitable magnetic or magneto-optical storage medium, such as resistive random access memory (DRAM), dynamic random access memory (DRAM), static random access memory (SRAM), enhanced dynamic random access memory (DRAM), high-bandwidth memory, hybrid memory cube, etc., or any other medium that can be used to store desired information and can be accessed by an application, module, or both. Any such computer storage medium can be part of a device or accessible to or connected to a device.

[0085] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A method for visually measuring the dissolution rate of water-soluble fertilizer, characterized in that, include: Obtain sequential images of fertilizer granules during the dissolution process; Taking any image in the sequence as the target image, perform multi-scale frequency domain analysis on each pixel in the target image to obtain the local phase and energy components at each scale. Calculate the weighted average phase using the energy components. Based on the phase deviation between the local phase and the weighted average phase, calculate the scale convergence index of each pixel. Divide all scales into a first scale group and a second scale group, obtain the energy response of each pixel in the target image under the first scale group and the second scale group, and calculate the spectral sharpness factor of each pixel based on the ratio of the energy response of the first scale group to the energy response of the second scale group. The solid boundary probability of each pixel is obtained by fusing the scale convergence index and the spectral sharpness factor; the target region of undissolved particles is segmented according to the solid boundary probability, and the physical mass of fertilizer is calculated based on the area of ​​the target region; each image in the sequence image is processed one by one to obtain the sequence of changes in physical mass over time, and the dissolution rate of fertilizer is calculated by differentiating the sequence of changes. Pixel Scale Convergence Index Satisfying the relation: ; It is a pixel. In the Energy amplitude at various scales; It is a pixel. In the Local phase at various scales; It is a pixel. The weighted average phase; It represents the total number of filter scales; It is the index value at the current scale; It is the preset first minute value; It is a refractive sensitivity control factor; It is a pixel. The standard deviation of the phase difference between the local phase and the weighted average phase at each scale; It is the hyperbolic tangent function.

2. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 1, characterized in that, The spectral sharpness factor of the pixel Satisfying the relation: ； in, It is a pixel. In the Energy amplitude at various scales; It is the sharpness sensitivity coefficient; It is the boundary index value between the first scale group and the second scale group; It is the preset second minute value; It is a natural exponential function.

3. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 1, characterized in that, The fusion scale convergence index and spectral sharpness factor are used to obtain the solid boundary probability of each pixel, including: A preset fusion index is introduced to perform exponential weighting on the spectral sharpness factor, resulting in a weighted spectral sharpness factor; The product of the scale convergence index and the weighted spectral sharpness factor is denoted as the solid boundary probability of a pixel.

4. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 1, characterized in that, The calculation of fertilizer physical quality based on the area of ​​the target region includes: An adaptive threshold segmentation algorithm is used to binarize the solid boundary probability map, which is composed of the solid boundary probabilities of all pixels in the target image, to obtain a fertilizer particle mask. Count the total number of pixels within the fertilizer particle mask; By combining camera calibration parameters, the total number of pixels is converted into physical size, and the physical mass of fertilizer particles is calculated based on an equivalent sphere model.

5. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 4, characterized in that, The dissolution rate of the fertilizer Satisfying the relation: ； in, It is about time Differentiation; It refers to the density of the fertilizer; yes The total number of pixels in the fertilizer particle mask in the time-series image; It is pi; It is the camera's pixel equivalent.

6. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 4, characterized in that, The adaptive threshold segmentation algorithm is the Otsu algorithm.

7. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 1, characterized in that, The calculation of the fertilizer dissolution rate includes: The remaining physical mass of the fertilizer is compared with the preset initial mass to obtain the percentage of physical mass. The moment when the physical mass percentage first falls below a preset residue threshold is defined as the complete dissolution time. If the complete dissolution time is less than the preset standard threshold, the current fertilizer is judged to have qualified dissolution performance.

8. The method for visually measuring the dissolution rate of water-soluble fertilizer according to claim 1, characterized in that, The multi-scale frequency domain analysis employs a logarithmic Gabor filter bank for multi-scale decomposition.

9. A visual measurement system for the dissolution rate of water-soluble fertilizer, characterized in that, The visual measurement system for the dissolution rate of water-soluble fertilizer includes a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a visual measurement method for the dissolution rate of water-soluble fertilizer according to any one of claims 1-8.

Images