An image detection method for fertilizer production

Abstract

The present invention relates to the technical field of image data processing, and particularly relates to an image detection method for fertilizer production, including: collecting multiple images during the composting process of organic fertilizers, calculating the straw point matching degree between each pixel point in the image and the surrounding neighborhood pixel points, and performing region growing based on the straw point matching degree to extract multiple straw regions; subsequently obtaining the area and gray mean value of each straw region, calculating the straw similarity index, and screening out the target region; further calculating the humification index of the target region of each image, and using the Mann-Kendall trend test method to calculate the value of the humification index of each image. If the value is less than the set threshold value, it is determined that the composting process of the organic fertilizer is completed. The present invention solves the problem of low accuracy in judging the composting process of organic fertilizers.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing. More specifically, the present invention relates to an image detection method for fertilizer production. Background Art

[0002] In modern agricultural production, the production process of organic fertilizers has received increasing attention, especially the quality monitoring and optimization of the composting process in fertilizer production. As a method for treating organic waste, composting can not only reduce environmental pollution, but also provide rich nutrients for the soil and improve soil quality. Therefore, how to efficiently and accurately monitor the changes in the composting process, especially the changes in the degree of humification, has become an important research direction in the field of fertilizer production.

[0003] Most traditional compost monitoring methods rely on manual sampling and laboratory analysis. This method not only takes a long time and is costly, but also is difficult to obtain dynamic data during the composting process in real time and is greatly affected by human factors. With the development of sensor technology, image processing technology, and computer vision technology, using image detection methods to monitor the composting process has gradually become a new trend. By regularly taking images of the composting area with a high-resolution camera, various changes during the composting process can be monitored in real time and comprehensively, especially the changes in the humification index during the composting process.

[0004] In addition, as the composting process progresses, the morphology, structure, and chemical composition of the composting materials will change, and traditional methods cannot reflect these changes in real time. Therefore, how to use image detection technology to evaluate each stage of the composting process in real time, especially the dynamic changes in the humification index, has become an important research direction for improving compost quality, shortening the composting cycle, and reducing resource waste. By combining image processing technology with modern data analysis means, not only can the monitoring cost be reduced, but also more accurate and real-time data support can be provided, thus promoting the automation and intelligentization of the organic fertilizer production process.

[0005] In the prior art, image processing technology is usually used to extract features from images and analyze the appearance changes of composting materials. However, most of these methods focus on the analysis of surface features such as the color and morphology of images, while ignoring the potential role of gray-scale information and texture information. In addition, image processing algorithms usually rely on simple threshold segmentation or edge detection techniques, which have high noise sensitivity and low efficiency problems, and cannot effectively extract the subtle changes during the composting process, resulting in the problem of low accuracy in judging the composting process of organic fertilizers. Summary of the Invention

[0006] To solve the problem of low accuracy in judging the composting process of organic fertilizers proposed in the above background art, the present invention provides the following solutions.

[0007] The present invention provides an image detection method for fertilizer production, including: collecting multiple images during the composting process of organic fertilizers; obtaining the straw point matching degree of each pixel point in the multiple images, where the straw point matching degree represents the probability that each pixel point belongs to the straw point area; performing region growing on each pixel point based on the straw point matching degree of each pixel point to obtain multiple straw regions, and obtaining the area of the minimum circumscribed rectangle of each straw region; obtaining the straw similarity index of each straw region, where the straw similarity index is positively correlated with the mean value of the gray values of the pixel points in the corresponding straw region and the area of the corresponding straw region, and is inversely correlated with the area of the minimum circumscribed rectangle of the corresponding straw region; taking the regions with the straw similarity index of each region greater than the set threshold as target regions; obtaining the humification index of the target regions of each image, where the humification index represents the humification degree of each image; using the Mann-Kendall trend test method to calculate the value of the humification index of each image, if the value is less than the set value threshold, then it is determined that the composting process of the organic fertilizer is completed.

[0008] Through calculating the matching degree of straw points in the images during the composting process of organic fertilizers and performing region growing, the above technical solution can accurately identify the straw regions and analyze their similarity, thereby effectively evaluating the humification degree during the composting process; by calculating the humification index of each image and applying the Mann-Kendall trend test method, it can objectively judge whether the composting process is completed, can quantitatively analyze the progress of composting in real time and automatically, provides a more efficient solution than the traditional manual method, and solves the problem of low accuracy in judging the composting process of organic fertilizers.

[0009] Further, the straw point matching degree is specifically: the straw point matching degree of the th pixel point, , where in the formula, is the gray value of the th pixel point in the set neighborhood centered on the th pixel point, is the gray value of the th pixel point, is a preset hyperparameter, is the number of pixel points in the set neighborhood centered on the th pixel point.

[0010] By introducing the calculation method of the straw point matching degree, the above technical solution introduces a smoothing adjustment factor while evaluating the gray similarity between a pixel point and its neighborhood pixels, can effectively suppress the influence of noise interference and local outliers on the region growing process, and thus improves the accuracy and stability of straw region extraction.

[0011] Further, the straw similarity index is specifically: the straw similarity index of the th straw region , , is the mean value of the gray values of the pixel points in the th straw region, is the area of the th straw region, is the area of the minimum bounding rectangle of the th straw region.

[0012] By constructing a straw similarity index that combines the internal gray mean value of the region, the ratio of the region area to the area of its minimum bounding rectangle, the above technical solution effectively synthesizes the brightness characteristics and morphological structure information of the region, making this index more comprehensive and accurate when measuring whether a region has typical straw characteristics. It can not only reflect the texture characteristics of the region but also suppress the interference of irregularly shaped or mis-extracted regions, thus improving the reliability and discriminability of straw region recognition.

[0013] Further, the humification index is specifically: the humification index of the th image is, , where is the entropy of the gray-level co-occurrence matrix corresponding to the th image, is the number of pixel points in the th target region in the th image, is the total number of target regions in the th image.

[0014] By introducing the entropy value of the gray-level co-occurrence matrix and the number of pixel points in the target region to calculate the humification index of the image, the above technical solution realizes a more comprehensive and accurate assessment of the humification degree in the composting process. It can reflect the complexity of the image texture through the entropy value and strengthen the quantitative analysis of the humification state of different regions by combining the number of pixels in the target region, thus improving the sensitivity and accuracy of monitoring. In this way, the progress of the organic fertilizer composting process can be judged more efficiently, providing reliable technical support for automated and real-time monitoring.

[0015] Further, a CCD camera is used to collect multiple images during the organic fertilizer composting process.

[0016] Further, the multiple images are subjected to denoising and grayscale processing.

[0017] The above technical solution can significantly improve the image quality by denoising and grayscaling multiple images, reduce the interference of noise on the analysis results, and thus ensure the accuracy of subsequent image processing and feature extraction. The denoising process helps to eliminate background noise and irrelevant interference information, making the effective features in the image more prominent. The grayscaling process simplifies the complexity of the image, facilitating a more accurate assessment of the degree of humification. These processing steps effectively improve the stability and reliability of image analysis, providing a clearer and more consistent image data basis for composting process monitoring.

[0018] Furthermore, the rotating calipers algorithm is used to obtain the area of the minimum bounding rectangle of each straw region.

[0019] The above technical solution can accurately determine the morphological characteristics of the straw region by using the rotating calipers algorithm to obtain the area of the minimum bounding rectangle of each straw region, overcoming the limitations of traditional methods in dealing with irregular shapes. The rotating calipers algorithm can accurately fit the outer boundary of the straw region and calculate the area of the minimum bounding rectangle, thus providing more accurate geometric features for subsequent similarity analysis and humification degree assessment.

[0020] Furthermore, the set value threshold is 0.9.

[0021] Furthermore, the obtaining of multiple straw regions includes: using the target pixel point as the seed point, where the target pixel point is any one of the pixel points. Sequentially judge whether the straw point matching degree of the neighborhood pixel points of the target pixel point is greater than the preset growth threshold, and incorporate the neighborhood pixel points greater than the preset growth threshold into the current growth region, and finally obtain multiple straw regions.

[0022] Furthermore, multiple images during the composting process of organic fertilizer are collected, and the time interval for image collection is set to be once every 12 hours.

[0023] The above technical solution can regularly monitor the changes of organic fertilizer during the composting process by setting the image collection time interval to be once every 12 hours, and capture the key stages in the composting process. This time interval can balance the frequency of image collection and the amount of data, ensure obtaining a sufficient time span to observe the humification process, and avoid data redundancy caused by overly frequent image collection. By collecting images at regular intervals, the dynamic changes of the composting process can be effectively tracked, providing representative data for subsequent analysis, and further improving the accuracy and timeliness of humification degree assessment.

[0024] The beneficial effects of the present invention are as follows:

[0025] By combining image acquisition, denoising processing, straw area recognition, and humification index calculation, the present invention realizes the precise monitoring of the organic fertilizer composting process. Through the accurate calculation of the straw point matching degree and the region growing algorithm, the straw area can be efficiently identified and analyzed, and the composting progress can be evaluated by combining the gray-level co-occurrence matrix of the image and the humification index. In addition, the application of bilateral filtering denoising and the rotating calipers algorithm improves the image quality and the accuracy of morphological recognition. Regularly collecting images and using the Mann-Kendall trend test method provide an objective quantitative basis for the completion of the composting process, making the monitoring of the fertilizer production process more intelligent, precise, and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 is a flowchart schematically showing an image detection method for fertilizer production according to an embodiment of the present invention;

[0027] Figure 2 is a gray-scale image of the straw composting process of an image detection method for fertilizer production according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] An embodiment of an image detection method for fertilizer production.

[0029] As Figure 1 shown, a flowchart of an image detection method for fertilizer production according to an embodiment of the present invention includes the following steps:

[0030] S1: Collect multiple images during the organic fertilizer composting process.

[0031] In one embodiment, a high-resolution CCD camera is used to collect images at multiple critical moments during the organic fertilizer composting process, thereby obtaining a series of image data with time sequence. The time interval of image collection can be set to once every 12 hours, and of course, it can also be set according to the actual situation; these image data reflect the details of the material changes during the composting process and can provide a visual basis for subsequent composting monitoring and quality control.

[0032] To improve the image quality, reduce the interference of noise, and ensure the accuracy and reliability of subsequent analysis, multiple collected images are first denoised and grayscale processed. During the denoising process, the bilateral filtering algorithm is selected as the main denoising method. Bilateral filtering is an effective image smoothing technique that can retain both the edge information and details in the image while removing the noise in the image. Different from traditional mean filtering or Gaussian filtering, bilateral filtering not only considers the spatial distance of pixels but also takes into account the color or intensity differences of pixels, enabling it to maximize the retention of image edges during the smoothing process and avoiding the problem of image blurring. When performing bilateral filtering denoising on grayscale images, the bilateral filter removes noise through the following two parts: on the one hand, it fuses neighboring pixels using the weights in the spatial domain; on the other hand, it weights according to the color differences or brightness differences of pixels, enabling pixels with similar colors or brightness to be retained, thus effectively reducing the impact of high-frequency noise and maintaining the details and edges of the image. In this way, the subtle changes during the composting process, such as the surface details of the materials and the transitional areas of the temperature gradient during the composting process, are clearly presented without being interfered by noise.

[0033] The grayscale processing converts the color image into a grayscale image, removing the color information, significantly simplifying the computational complexity of subsequent image processing, and being able to focus on the brightness changes of the image. This processing method effectively removes color interference, making the key features of the image more prominent. Especially when analyzing the morphology, distribution, and surface changes of the materials during the composting process, it can provide a clearer perspective. Grayscale images are particularly important for detecting the degree of maturity of compost materials because they can better reflect the brightness differences in different regions, helping to distinguish the different states of compost materials in detail. In addition, grayscale images can effectively reduce the impact of image noise, improve the stability and accuracy of subsequent image processing algorithms, and provide a more efficient and accurate basis for the automated monitoring and evaluation of the composting process.

[0034] As Figure 2 shown, the grayscale image of the straw composting process of an image detection method for fertilizer production according to an embodiment of the present invention includes: four processes of early straw composting, middle straw composting, late straw composting, and straw composting maturity.

[0035] S2: Calculate the straw point matching degrees of each pixel point in multiple images, and perform region growing on the straw point matching degrees of each pixel point to obtain multiple straw regions.

[0036] In one embodiment, calculate the straw point matching degree of the th pixel point in each image , , where is the th pixel point as the center within a set neighborhood, and the The grayscale value of a pixel is the grayscale value of the pixel, is a preset hyperparameter, and is the number of pixels within a set neighborhood centered on the pixel;

[0037] By calculating the straw point matching degree of each pixel in each image, the aim is to evaluate the similarity between a certain pixel in the image and other pixels within its neighborhood. The larger the grayscale value of the pixel, the whiter the corresponding color, the less likely it is to be a background area, and the more in line with the color characteristics within the straw area; at the same time, the smaller the difference in grayscale values between the pixel and its neighborhood pixels, the less likely it is to be a straw edge area, and the more in line with the structural characteristics within the straw area. The more the color characteristics and structural characteristics of the pixel and its neighborhood pixels match the color characteristics and structural characteristics within the straw area, the more likely the pixel is to be a pixel within the straw area.

[0038] The set neighborhood can be an 8-neighborhood, and the preset hyperparameter can be 1. Of course, it can also be set according to the actual situation.

[0039] Perform region growing on the straw point matching degrees of each pixel to obtain multiple straw areas, and obtain the area of the minimum bounding rectangle of each straw area;

[0040] The obtaining of multiple straw areas includes: taking a target pixel as a seed point, where the target pixel is any pixel among the above-mentioned pixels, and sequentially determining whether the straw point matching degree of the neighborhood pixels of the target pixel is greater than a preset growth threshold, and incorporating the neighborhood pixels with a straw point matching degree greater than the preset growth threshold into the target pixel area, finally obtaining multiple straw areas.

[0041] The region growing is a prior art and will not be elaborated in this solution.

[0042] S3: Calculate the straw similarity index of each straw area, take the areas with a straw similarity index greater than a set threshold in each region as target regions, and calculate the humification index of the target regions in each image.

[0043] In one embodiment, calculate the straw similarity index of the straw area, , , where is the average value of the grayscale values of the pixels within the straw area, is the area of the straw area, which can be represented by the number of pixels, The area of the minimum bounding rectangle of each straw region, and the area of the minimum bounding rectangle of each straw region is obtained by using the rotating calipers algorithm; reflects the straw structure matching degree of the

[0044] th straw region, that is, the closer the shape structure of the straw is to a rectangle, the higher the matching degree with the straw structure;

[0045] Regard the regions where the straw similarity index of each region is greater than the set threshold as the target regions;

[0046] The humification index of the th image is , where is the entropy of the gray-level co-occurrence matrix corresponding to the th image, is the number of pixel points in the th target region in the th image, is the total number of target regions in the th image. During the composting process of straw, large pieces of straw are gradually decomposed into multiple small pieces, so the number of pixel points in the straw region will gradually decrease, that is, will gradually decrease over time, and the corresponding degree of maturity will gradually increase; at the same time, the greater the entropy of the gray-level co-occurrence matrix corresponding to the entire composting image, the more complex the texture structure in the composting image, corresponding to the texture characteristics of the fiber structure being damaged and the edges being rough after the straw is gradually decomposed, so the corresponding degree of maturity is greater, and thus the calculated humification index is greater.

[0047] By dynamically calculating the humification index, not only can the decomposition process of straw be accurately tracked, but also the change in the degree of straw maturity during the composting process can be better reflected, providing a scientific basis for the monitoring and evaluation of compost quality.

[0048] S4: Use the Mann-Kendall trend test method to calculate the value of the humification index of each image. If the value is less than the set value threshold, it is determined that the composting process of the organic fertilizer has been completed.

[0049] In one embodiment, the Mann-Kendall trend test method is used to analyze the humification index of each image, aiming to evaluate the completion degree of the organic fertilizer composting process. First of all, the Mann-Kendall trend test method, as a non-parametric statistical method, is widely used in the trend analysis of time series data. It can effectively identify the monotonic change trend in time series data without relying on the distribution assumption of the data. Therefore, it has obvious advantages when dealing with the dynamic changes of the humification index in the composting process, which may show a non-linear trend.

[0050] By combining the humification index corresponding to each image with the time series, the Mann-Kendall test can evaluate the trend of the humification index changing with time and calculate a value for each image. The value reflects whether the humification index shows a significant change trend. The smaller the value means that the change trend of the humification index in the composting process is more obvious, that is, the progress of the composting process is more significant. Correspondingly, if the calculated value is less than the preset value threshold, it indicates that the humification index in the image has changed significantly, which usually means that the composting process of the organic fertilizer is approaching completion.

[0051] Specifically, the humification index in the composting process gradually increases over time, reflecting the process of straw gradually decomposing and the degree of maturity gradually increasing. When the change trend of the humification index reaches the statistically significant level, that is, the value is less than the set threshold, it indicates that the decomposition of substances in the composting process has entered the final stage, meeting the standard of composting completion. Therefore, through the Mann-Kendall trend test method, it is possible to accurately judge whether the composting process has been completed, thereby providing a scientific basis for the production process of organic fertilizers and ensuring that the composting effect reaches the expected goal.

[0052] The solution of the present invention effectively improves the monitoring accuracy of the humification degree in the composting process by accurately calculating the straw point matching degree in the image, using the region growing algorithm to identify the straw region and evaluate its similarity. Combining the calculation of the gray level co-occurrence matrix entropy value and the humification index can quantitatively reflect the changes in the composting process and provide a scientific basis for the production process. Using the bilateral filtering denoising algorithm and the rotating calipers algorithm ensures the accuracy of image quality and morphological analysis, avoids the influence of noise, and can accurately fit the outer boundary of the straw region. By setting an appropriate time interval for image acquisition and real-time tracking of the dynamic changes in the composting process, the timeliness and reliability of the humification process evaluation are further improved, providing strong technical support for the automated monitoring and optimization of the fertilizer production process.

[0053] In the description of this specification, "a plurality of" and "several" mean at least two, such as two, three or more, etc., unless otherwise specifically defined.

[0054] Although this specification has shown and described multiple embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will think of many changes, alterations and alternative ways without departing from the spirit and scope of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in the practice of the present invention.

Claims

1. An image detection method for fertilizer production, characterized in that, Including: Collecting multiple images during the composting process of organic fertilizers; Obtain the straw point matching degrees of each pixel point in the multiple images, where the straw point matching degree represents the probability that each pixel point belongs to the straw point region, and the straw point matching degree of the th pixel point is: , where is the gray value of the -th pixel within a set neighborhood centered on the -th pixel point, is the gray value of the -th pixel point, is a preset hyperparameter, is the number of pixel points within a set neighborhood centered on the -th pixel point; Performing region growing on each pixel point based on the straw point matching degree of each pixel point to obtain multiple straw regions, and obtaining the area of the minimum circumscribed rectangle of each straw region; Obtain the straw similarity index for each straw area. The straw similarity index of the th straw area is as follows: , is the mean of the gray values of the pixel points in the th straw area, is the area of the th straw area, is the area of the minimum circumscribed rectangle of the th straw area; Regard the area where the straw similarity index of each straw area is greater than the set threshold as the target area; obtain the humification index of each image target area, where the humification index characterizes the humification degree of each image. The humification index of the th image is: , where is the entropy of the gray-level co-occurrence matrix corresponding to the th image, is the number of pixel points in the th image within the th target region, is the total number of target regions in the th image; Calculate the value of the humification index of each image using the Mann-Kendall trend test method. If the value is less than the set value threshold, it is determined that the organic fertilizer composting process is completed.

2. The image detection method for fertilizer production according to claim 1, wherein, Using a CCD camera to collect multiple images during the composting process of organic fertilizers.

3. The image detection method for fertilizer production according to claim 1, wherein, Performing denoising and grayscale processing on the multiple images.

4. The image detection method for fertilizer production according to claim 1, characterized in that, Using the rotating calipers algorithm to obtain the area of the minimum circumscribed rectangle of each straw region.

5. The image detection method for fertilizer production according to claim 1, characterized in that, The setting The value threshold is 0.

9.

6. The image detection method for fertilizer production according to claim 1, wherein The obtaining of multiple straw regions includes: using a target pixel point as a seed point, the target pixel point being any one of the pixel points, sequentially determining whether the straw point matching degree of the neighborhood pixel points of the target pixel point is greater than a preset growth threshold, incorporating the neighborhood pixel points with a straw point matching degree greater than the preset growth threshold into the target pixel point region, and finally obtaining multiple straw regions.

7. The image detection method for fertilizer production according to claim 1, wherein, For the collecting of multiple images during the composting process of organic fertilizers, the time interval for image collection is set to be once every 12 hours.

Images